Of the direct debit collections on Adfin that suffered a failed attempt, 60.1% are still recovered (Adfin platform data). Look at the attempts themselves and you can see where that recovery comes from: a second presentation succeeds 46.9% of the time, a third 25.0%, and a fourth 19.1%. So the first fortnight after a failure is where most of your money is, and an automatic re-presentation is what collects it.
In this article
The short version
- 97.0% of direct debit collections succeed at the first attempt and 98.7% are collected in the end (Adfin platform data).
- Only 2.6% ever see a failed attempt, and 60.1% of those are recovered anyway.
- A second attempt succeeds 46.9% of the time. A third drops to 25.0% and a fourth to 19.1%.
- Your attempts run a median of 6 days apart. That's the Bacs re-presentation cycle and not a tuned interval.
- Most returned collections are mechanical, so the reason code tells you whether another attempt has any chance.
Two different cuts of the same data
Before any of the numbers, a warning about mixing them, because the two figures you'll see here are calculated on different denominators and reading them as one series would give you nonsense.
The 60.1% in the title is the share of collections that suffered a failed attempt and were nonetheless recovered. Its denominator is the collections that failed at least once. The table below is calculated differently: it takes the attempts made at each stage of the sequence and gives the share of those attempts that succeeded, so its denominator is attempts and not collections.
Both are true and they measure different things. A collection can be recovered on its second attempt, its third or by another route entirely, and that's why 60.1% of them come back to you while no single later attempt gets anywhere near that rate on its own.
The success rate at each attempt
Here's what each attempt in the sequence collects, taken across every direct debit attempt on Adfin (Adfin platform data).
| Attempt | Succeeded | Median days since the previous attempt |
|---|---|---|
| 1st | 97.0% | n/a |
| 2nd | 46.9% | 6 |
| 3rd | 25.0% | 6 |
| 4th | 19.1% | 6 |
| 5th | 16.4% | 6 |
Adfin platform data. Each row is the share of the attempts made at that stage that succeeded, measured on 269,518 direct debit collections first attempted between July 2024 and mid-July 2026.
Two things to take from that. Almost everything settles first time, so the population you're making decisions about is small, at 2.6% of your collections. And the value of retrying is concentrated in the attempt straight after the failure, where it's close to a coin flip. After that the odds halve and then flatten out in the high teens.
So the money is in presenting again at all, and the common mistakes lie either side of that. A collection nobody re-presents gives up a 46.9% chance of your money for nothing. A sequence that keeps presenting against a cancelled instruction spends your customer's patience on an attempt with no mechanism behind it. The likely reading of the decay is that collections still failing that late are increasingly the ones where the arrangement itself has gone, although nothing in the data records why any single attempt failed.
Why a second attempt works
The reason is in what a returned collection usually means. Bacs describes ARUDD as "the service allowing PSPs to return Direct Debit payments to the service user that they could not apply", and every return arrives with a reason code attached.
Bacs doesn't publish the code lists openly, so what follows is the consensus across several provider reproductions, not a scheme statement. Their ARUDD lists agree on codes for an empty balance, a cancelled instruction, a transferred or closed account, a disputed advance notice, a wrong account type, an amount that differs from what was notified and an amount not yet due. Read down that list and most of it describes an administrative state on a particular day, not a decision your customer made about you.
An empty balance on the 1st is often a full one on the 8th. A payer who has moved bank hasn't refused to pay you, they've just left your instruction behind. An amount that didn't match the notice you sent needs a corrected notice more than it needs a conversation. Time and a second attempt fix a lot of it without you negotiating anything, and that's the 46.9%. The decay after it is the other half of the same story, because a balance can fill up while a closed account never will.
There's a second effect worth separating out. Some of what looks like a fourth or fifth attempt is a collection that changed shape along the way, because the customer paid another way, or the mandate was fixed and the collection re-presented. So the sequence isn't purely the same instruction hitting the same wall repeatedly.
What stopping at the first failure costs you
Think about what a write-off rule actually does. If you treat the first failure as the end of it, you're giving up on an attempt that succeeds 46.9% of the time, on collections where nothing has happened except a date landing in the wrong week. If you stop after two, you're walking away from a 25.0% chance, and after three from 19.1%.
The amounts involved are usually small, and that's precisely why the rule persists. Chasing a £180 fee through a fourth attempt feels uneconomic when someone on your team has to do it by hand. Failures run at 2.6% of collections, a low enough rate that nobody builds a process for it, so the work goes to whoever happens to notice.
What changes the calculation is who does the retrying. If a person has to remember, an early cutoff is a rational response to their time. If the sequence runs on its own and costs you nothing per attempt, then a 19.1% chance on the fourth is still worth taking, because the alternative is a written-off invoice.
There's a cost on the other side too. A provider that charges you for each failed collection or each retry is pricing you towards giving up early, and the fee is generally small next to the collection you abandon because of it.
When to stop retrying
Not every failure deserves another attempt, and your reason code tells you which is which. Again, these code meanings come from provider documentation, since no published scheme list exists.
- Insufficient funds or refer to payer: retry, ideally a few days later and not the next morning.
- Instruction cancelled, or no instruction held: stop retrying and get a new mandate signed, because further attempts against a dead instruction go nowhere.
- Account transferred or closed: stop, ask for current details, and set up a fresh mandate.
- Advance notice disputed, or amount differs from the notice: fix the notice before you present again, because repeating the same collection repeats the same objection.
- Payer deceased: stop, and handle it as an estate matter.
Retrying blindly against codes 2 to 5 is what gives retries a bad name, and it's a plausible explanation for the fourth and fifth attempts collecting as little as they do. The figures above describe sequences where the response matched the reason, so if your sequence ignores the code it won't reproduce them.
Designing a retry sequence worth running
Four things make the difference between a sequence that recovers money and one that just annoys people.
Space the attempts. A balance that was empty this morning is unlikely to be full this afternoon. Attempts on Adfin run a median of 6 days apart, which is the Bacs re-presentation cycle rather than a tuned gap, so treat that as the shape of the sequence and not as evidence that 6 days beats 5.
Give notice each time. The Direct Debit Guarantee commits you to telling the payer about changes to the amount, date or frequency, "normally 10 working days" in advance "or as otherwise agreed", and a re-presentation your customer knew about is more likely to clear.
Keep another route open. A card or a payment link can settle the same invoice the same day, and that saves you waiting for the next collection date on an account that's currently empty.
Make sure the attempts cost you nothing. Adfin doesn't charge you for retries and can fall back to card automatically when a direct debit fails, so a third attempt at 25.0% stays worth having instead of becoming a decision somebody has to justify.
Beyond that, put a person on the small number that survive your sequence. By then you've filtered out the empty balances and the stale account details, so what's left in front of you is far more likely to be a customer who genuinely needs a conversation.
Common questions
Do failed direct debits usually get paid in the end? Most do. On Adfin, 97.0% of direct debit collections succeed at the first attempt and 98.7% are collected eventually. Of the 2.6% that see a failed attempt, 60.1% are recovered anyway.
Is it worth retrying a direct debit four or five times? The attempt straight after a failure certainly is, at 46.9%. A third succeeds 25.0% of the time and a fourth 19.1%, so later attempts earn their place where the reason code allows them and each one costs you nothing to run.
What does the 60% figure actually mean? It's the share of direct debit collections that suffered a failed attempt and were still recovered, at 60.1%. That's a different calculation from the per-attempt figures. Those give the share of the attempts made at each stage that succeeded, so the two shouldn't be read as one series.
When should you stop retrying? When the reason code tells you the instruction or the account has gone. A cancelled mandate, no instruction held, a closed or transferred account and a deceased payer all need a new arrangement instead of another attempt.
How long should you leave between attempts? A few days, not a day. Insufficient funds is a timing problem, so an attempt spaced to land after your customer's own pay cycle has a better chance than one the next morning.
Does retrying annoy customers? Less than you'd expect when each attempt carries advance notice and the reason code is respected. What tends to cause complaints is repeating a collection your customer has already disputed, or presenting against a mandate they cancelled.
Sources
- Bacs — glossary (accurate as of August 2026)
- Direct Debit — the Direct Debit Guarantee (accurate as of August 2026)
This article describes retry outcomes on Adfin's own platform and is not legal or financial advice. Reason code meanings are taken from provider documentation, since Bacs does not publish the code lists openly. Last updated August 2026.
