Product

RECOMMENDED

FREE TRIAL

Resources

Q3 2026 Edition

Baremetrics Recovery Index: the payment recovery benchmark for subscription businesses

203,756 failed payments34,252 payments recovered457,225 dunning emails

How often failed subscription payments come back, how quickly, and which dunning emails bring them back, with a checker to see where your own rate falls.1

Check your recovery rate

Pick your recovery rate and company size to see where you fall among similar companies in this edition. Nothing you enter is sent or stored.

Your exact recovery rate is already calculated in Baremetrics Recover. See your real recovery rate

01What the typical company recovers

The typical company recovered 29.4% of the failed charges its dunning system attempted between July and September 2026. That median is this edition's Recovery Index.

29.4%Recovery Index: median attempted recovery rate across all companies in this edition
Under $50K45% of companies $50K to $250K38% of companies $250K or more17% of companies MRR band Median 29.4% You 0%10%20%30%40%50%60%70%80% Attempted recovery rate
34,252charges recovered
203,756charges attempted
$3,690,677revenue recovered
How this is measured

The Recovery Index is the median attempted recovery rate: recovered charges divided by the failed charges Recover actively tried to recover, calculated for each company, then taking the company in the middle of the sample. Each size group is shown as 50 dots, and each dot stands for 2% of that group's companies, placed at that slice's recovery rate. Dots don't correspond to individual companies. The pooled rate, 16.8%, divides all recovered charges in the sample by all attempted charges.

Compare your own rate with the index, not the pooled rate. The pooled rate is pulled down by a small number of very large senders, as section 07 shows.

02How this benchmark was built

One quarter of real Recover data, measured the same way every edition.

Who's included
Every Baremetrics Recover customer with at least one attempted recovery in the quarter. Nobody was hand-picked or left out for poor results.
Time period
July 1 to September 30, 2026
Where the numbers come from
Headline figures come from the daily metrics customers see in their Recover dashboard. Timing and email figures come from Recover's message records.
Privacy
No company is named, no chart shows an individual company, and company counts aren't published.
Data as of
November 18, 2026

Read the full methodology

03How recovery rates vary between companies

Every size group includes companies that recover almost nothing and companies that recover well over half. What changes is where the middle half sits: highest for $50K to $250K in MRR, lowest for the largest companies.

All companies All size groups
Box plot of attempted recovery rates for the selected size group. All companies 29.4% You 0%10%20%30%40%50%60%70%80% Attempted recovery rate
How this is measured

The box shows the middle half of companies in the selected group, the thick line inside is the median, and the whiskers run from the lowest to the highest rate. Each dot stands for 2% of the companies in the group, placed at that slice's recovery rate; dots don't correspond to individual companies. When you pick a size group, the dashed line marks the median for all companies.

The lowest-recovering tenth of companies recover 6.3% or less; the highest-recovering tenth recover 55.5% or more.

A gap this wide suggests there's more to recovery rate than the tool itself. Your sequence, customer base and billing model matter as much as having dunning switched on.

How to improve your dunning recovery rateRead the guide

04How quickly failed payments come back

Half of recovered payments are paid within three full days of failing, and 90% within two weeks.

Paid by day i
52% of recovered payments were paid within three days of failing 100%75%50%25%0% Day 0Day 7Day 14Day 21Day 30
Share of recovered payments paid by each day after the payment failed, July to September 2026. Hover the line to read any day.
How this is measured

Days count full days after the first failed attempt; day 0 is the same day. Every failed payment that was later paid counts, whether or not an email was opened first.

Payments jump on days 14, 21 and 28 rather than tapering smoothly; day 14 saw more than twice as many as day 13.

The first few days after a failure carry most of the opportunity. A slow first email costs more than a missing last one.

Why subscription payments fail, and how to fix each causeRead the article

05One in six recoveries needs no email

17% of recovered payments were paid before the first dunning email went out, for example because the payment processor retried the card or the customer updated their details on their own. No email prompted these recoveries, so this report counts them separately.

All recovered payments, July to September 2026

Hover either part of the bar for detail.

How this is measured

A payment counts as recovered before any email when it was paid before the first dunning email for it was sent.

Counting these as dunning wins makes your emails look about 20% more effective than they are. For payments paid after the first email went out, the median came four days after the failure.

The complete guide to involuntary churnRead the guide

06Which emails in the sequence do the work

The first three dunning emails, sent on days 0, 3 and 7, account for 89% of recoveries that follow an email. The day 7 email recovers the most per send, with day 3 close behind.

    Later emails reach far fewer payments, since most have already been recovered or abandoned by then, so their lower rates partly reflect who is still unpaid.

    How this is measured

    Each recovery is credited to the last email sent before the payment came back. Only failed-payment emails count, not card-expiry or renewal reminders.

    The first week's emails deserve most of your attention. The later steps still bring back about one recovery in nine, so they're worth keeping, but they aren't the main lever.

    How to write dunning emails that convertRead the guide

    07The biggest companies send most of the volume

    A small group of large companies accounts for most of the recovery attempts in this report. Companies with $250K or more in MRR are about one in six companies here, but more than two-thirds of the failed charges Recover tried to recover were theirs. They also recover less than most, which pulls the overall rate down to 16.8%, even though the typical company recovers 29.4%.

    Share of companies compared with share of attempted charges
    Share of companies
    Share of attempted charges203,756 charges

    Hover any segment for detail.

    Mid-sized companies are the opposite. They make up 38% of companies but only 22% of attempted charges, and they recover more of them than any other group (29.9%). This shows the pattern, not the reason behind it.

    How this is measured

    Each company's MRR is its average over the quarter, in its own reporting currency; about 91% of the sample reports in US dollars. Share of attempted charges is each group's attempted recovery charges divided by all 203,756 attempted in the quarter. The overall (pooled) rate divides all recovered charges by all attempted charges, so the biggest senders count the most.

    To see how you're doing, compare yourself with the typical company (29.4%) or with companies your size in section 03. The overall rate mostly tells you what the biggest senders are doing.

    The complete dunning management guideRead the guide

    08How this quarter compares with May 2026

    The median company's monthly recovery rate has stayed between 29% and 34% every month since May.

    Monthly recovery rate, May to September 2026 40%20%0% Median companyPooled MayJuneJulyAugustSeptember
    Hover the chart to read each month. September is dashed because its figures are still rising as late recoveries are credited.

    Monthly rates aren't directly comparable with the quarterly index in section 01, since each company's rate is calculated over a shorter period.

    Restated May 2026 figures: our first edition reported a 12.7% median for May 2026. On current data, with every Recover customer included, May's median is 34.0%. The full explanation is in the methodology.

    09What to take from this

    • 89%

      Get the first week right: The emails on days 0, 3 and 7 account for 89% of recoveries that follow an email, and the day 7 email recovers the most per send, with day 3 close behind. Most of what dunning can win back is won in the first week.

    • 17%

      Don't credit dunning for everything: About one in six recoveries happens before any email goes out, for example through processor retries or customers updating their card. Measure attempted recovery rate so you can see what your emails actually add.

    • 40%

      Compare like with like: Mid-sized companies ($50K to $250K MRR) recover a median 40% of attempted charges, about twice the 19.5% of the largest companies. Use your size band as your benchmark, not the overall 29.4%.

    • 1 in 9

      Keep the later emails, but don't expect much from them: Steps after day 7 bring back about one recovery in nine. They're a backstop, not the main lever.

    FAQ

    Questions

    What is the Baremetrics Recovery Index?

    The Baremetrics Recovery Index is the median attempted recovery rate across subscription businesses using Baremetrics Recover, published every quarter. For July to September 2026 it was 29.4%, meaning the typical company recovered 29.4% of the failed charges its dunning system attempted.

    What is a good dunning recovery rate?

    Across subscription businesses using Baremetrics Recover in July to September 2026, the median attempted recovery rate was 29.4%, and the middle half of companies recovered between 16.0% and 47.5%. A good rate depends on company size: mid-sized companies ($50K to $250K MRR) had a median of 39.9%.

    How quickly are failed subscription payments recovered?

    Half of recovered payments were paid within three full days of failing, three-quarters within a week and 90% within two weeks.

    Which dunning email recovers the most payments?

    The first three emails, sent on days 0, 3 and 7, accounted for 89% of recoveries that followed an email. The day 7 email had the highest recoveries per send, at 15.3 per 100, just ahead of day 3 at 14.9.

    Do failed payments recover without dunning emails?

    Some do. About one in six recovered payments (17%) were paid before any dunning email was sent, for example through processor retries or customers updating their card.

    Why is the median recovery rate higher than the overall rate?

    The largest companies send most of the volume and recover at lower rates. Companies with $250K or more in MRR made up 68% of attempted charges and had a median rate of 19.5%, which pulls the pooled rate down to 16.8%.

    Why are the May 2026 figures different from the first report?

    Recover sometimes credits a recovery days or weeks after the payment, dated to the payment. More May recoveries have been credited since the first report, and this edition includes every Recover customer with an attempted recovery rather than a subset. The restated May median is 34.0%; the original 12.7% couldn't be reproduced from current data.

    Methodology

    This report covers subscription businesses using Baremetrics Recover between July 1 and September 30, 2026. Every figure comes from Recover's own data, measured with the same definitions customers see in their Recover dashboard.

    The sample

    The sample is every Baremetrics customer that had Recover active and at least one attempted recovery during the quarter. Nobody was hand-picked or left out for poor results.

    We don't filter by country, industry or business model, so the results describe subscription businesses using Recover, not any one segment. To protect customer confidentiality, we don't publish the number of companies in the sample or in any group; group sizes appear only as shares. No company is named, and charts that show the spread of rates use dots that each stand for 2% of a group's companies, not individual companies.

    Where the numbers come from

    The headline figures come from the recovery metrics Recover calculates for each customer every day. These are the numbers shown in each customer's Recover dashboard, summed over the quarter. We checked them against Recover's raw dunning records, and they match.

    The sections on payment timing and sequence steps use Recover's message records instead, because only these link each failed payment to the emails sent about it. They count failed-payment emails only, not card-expiry or renewal reminders. The message records count each failed payment once and include payments made before any email was sent, so their totals are larger than the headline recovered charges. Those sections report shares and percentages, not totals.

    What we measure

    Attempted recovery charges
    Failed charges that Recover actively tried to recover. One-time charges and first-time failures that payment providers often cancel immediately are excluded.
    Recovered charges
    Attempted recovery charges that were later paid.
    Attempted recovery rate
    Recovered charges divided by attempted recovery charges, calculated for each company.
    Recovery Index
    The median attempted recovery rate across the sample, calculated the same way every quarter.
    Pooled recovery rate
    All recovered charges in the sample divided by all attempted recovery charges.
    Recovered revenue
    The value of recovered charges in US dollars, with each charge converted at the exchange rate on its charge date.
    Days to payment
    Full days between a payment's first failure and the day it was paid; day 0 is the same day.
    Company size
    Each company's average MRR over the quarter, in its own reporting currency, grouped as under $50K, $50K to $250K, and $250K or more. About 91% of the sample reports in US dollars.

    Attempted recovery rate is the primary metric because it only counts failures the dunning system actually worked on. A rate based on all failed charges also counts payments a company chose not to pursue, such as manually invoiced accounts or customers in a payment arrangement. That makes dunning look worse than it is and makes companies impossible to compare.

    How results are summarized

    We report the median company and the middle half of companies (25th to 75th percentile) as the main figures. A few companies with very high failure volumes recover at lower rates, which pulls the pooled rate well below the typical company's. The median is the right comparison for your own business; the pooled rate shows where most of the sample's failed charges were.

    In the sequence step section, each recovery is credited to the last email sent before the payment came back. Recoveries that happen before any email is sent are reported separately and aren't credited to dunning.

    Restated May 2026 figures

    Our first benchmark covered May 2026, using Recover's metrics as they stood when it was prepared in June 2026. More May recoveries have been added since. Recover sometimes detects a recovery days or weeks after the payment and credits it to the payment's date. This edition also counts every Recover customer with an attempted recovery, where the first covered a subset of customers. The restated recovered charges and revenue are counted from Recover's raw records, because the stored metrics no longer include an account that has since left Recover.

    May 2026As first publishedRestated
    Recovered charges10,99913,142
    Recovered revenue$1,236,764$1,331,455
    Median attempted recovery rate12.7%34.0%

    We use the restated figures throughout this edition so every month is measured from the same data. The original median of 12.7% could not be reproduced from the current data. Comparisons with the first edition should use the restated figures.

    Limitations

    Every company in the sample chose to pay for a dedicated recovery tool. They may care more about failed payments than the average subscription business, or have more of them, so these results shouldn't be read as typical for companies relying on processor defaults.

    Recent figures can still rise. Recoveries detected late are credited to the payment's date, so a month's figures can grow after it ends. All figures here are as recorded on November 18, 2026.

    About 95% of companies with Recover activity bill through Stripe, a few alongside Braintree; the rest bill through Braintree alone. That reflects our customer base, not the wider market, and results by billing system aren't broken out.

    Sequence steps are labelled by each email's current scheduled day. If a company changed its sequence during the quarter, some recoveries may be credited to the wrong step.

    Who produced this

    Baremetrics produced this report from Recover customer data. We sell a payment recovery product, so we have an interest in the results. That's why the method and definitions are published in full, so you can judge the numbers yourself.

    How this benchmark is updated

    This benchmark is updated quarterly on this page, using the same definitions and sample rule each time. If a definition changes, we'll say what changed and restate earlier figures where we can.

    Past editions: May 2026 (restated), July to September 2026