Group your clients by how they've actually paid you before, then vary three things per group: how you collect, when the first chase goes out, and how far the escalation runs. Payment history sorts a ledger better than invoice size or sector does, and you can build the groups from your own records in an afternoon.
In this article
The short version
- Among repeat payers in Adfin's book, 27.4% were never late once, while the 24.3% late most of the time produced 51.9% of every late payment (Adfin platform data, 12,863 payments across 1,359 business-customer pairs with at least five payments each).
- Those pairs qualified by paying five times themselves, so they're a selected group: use the shape and not the proportions.
- Build your groups from one number per client, the share of their invoices that arrived after the due date.
- Vary how you collect and how far you escalate, and keep the wording identical.
- Nobody can give you the optimal number of reminders per group, us included, because our warehouse holds no chasing-event data to derive it from.
Why payment history is the split worth using
Payment history takes more work to assemble than invoice value or sector does, and it tells you more. Across 1,359 business-customer pairs with at least five payments the customer made themselves, covering 12,863 payments, 27.4% of pairs were never late once, 10.3% were late on every invoice, and the worst 24.3% accounted for 51.9% of all late payments (Adfin platform data). On a two hundred client ledger, that shape puts about fifty names behind half your chasing work.
One limit travels with it. A pair only enters that cut after five customer-initiated payments, and a client who paid promptly by card five times running is exactly the one most firms would have moved onto a mandate by then. So the sample leans towards clients who resisted, and the proportions won't transfer to your own book.
Invoice size does worse as a split. Among late payments the median delay is 8 days in every size band, and inside one business's own book 39 businesses ran later on their large invoices against 43 later on their small ones (Adfin platform data). Big invoices need care for other reasons and don't earn their own group.
Four groups you can build from your own ledger
Four groups is usually enough, plus a holding pen for clients you don't know yet.
| Group | How you spot it | What you do differently |
|---|---|---|
| Pays every time | never late across five or more invoices | invoice and a due-date reminder, nothing after it |
| Slips by a few days | late sometimes, almost always inside a week | scheduled reminders, no calls, no interest |
| Late most months | late on around half their invoices or more | change how you collect: mandate, instalments, payment up front |
| Late and quiet | late and not replying, or a balance ageing past 60 days | named owner, statements, work paused, interest and formal notice |
| No history yet | fewer than five invoices with you | your default rules until a record exists |
Pick your thresholds and write them down. Ours: nothing overdue for the first group, up to about 40% of invoices late for the second, above that for the third, and anything unanswered for 30 days jumping to the fourth.
What changes per group, and what stays the same
Your wording can stay the same for every group. These are the things worth varying:
- How you collect, which carries more weight than anything you write. A client late most months is asking for a mandate or an instalment plan, so their invoice stops depending on them remembering.
- Your terms, at the next fee review. Shorter terms, a deposit, or payment before the next piece of work.
- When your first contact goes out. Your reliable group doesn't need a reminder on day one overdue, and your persistent group needs a confirmation before the due date instead.
- Who owns it. Your fourth group needs a named person and a diarised call, and your first two need nobody.
- Where you stop. Interest, a letter before action and a hold on new work belong to the bottom group, in the order set out in recovering overdue debt.
Differentiated tone reads as inconsistency the first time two clients compare notes, so your reminder templates can be identical across all four groups while the schedule and the mechanism differ completely.
Setting it up on Monday
- Export twelve months of paid invoices with the client, the due date and the date the money arrived. Your accounting software will do it, and an aged debtor report covers today.
- Work out days late per invoice from the due date and the payment date. Try not to treat a binary overdue flag as days, because a flag says whether an invoice went overdue and never how far.
- Handle direct debit separately, because Bacs settles a few days after the due date by design. Those invoices read as late when nothing has gone wrong, so mixing the two populations puts half your clients in the wrong group. Why direct debit payments look late explains the cycle.
- Per client, count their invoices and the share that arrived after the due date. Fewer than five and they go into your default group.
- Rank the rest, cut at your thresholds, and read the names. Where somebody lands surprisingly low, your data is probably right and your impression came from one phone call.
- Write the rules on one page, against the group and never against a client name.
- Tag the group in your ledger or your credit control tool so the rule fires without you deciding, then re-run the cut quarterly.
What no data can tell you about cadence
You'll want to know how many reminders each group needs and how far apart. No defensible answer is available, including from us: our warehouse records when invoices were paid and how, with no record of what was sent when, so no optimal cadence per group comes out of it.
What holds up is narrower. Decide each group's schedule before the invoices go out, keep it predictable, and change the medium instead of repeating an email. How often to chase an unpaid invoice covers the default rhythm your groups vary.
Clients move between groups
Your quarterly re-run matters more than the initial sort, and it works both ways: somebody who has slipped three months running has told you something, and somebody who has paid six invoices on time since a bad spell has earned their way back out. A group people can only fall into becomes a punishment list.
When holding the rules shouldn't be your job
The rule here is easy to write and hard to run. Vary how you chase by what each client has done before, review it quarterly, keep your promises consistent. You can hold that across ten accounts, and across two hundred it tends to last until the first busy week. Businesses affected by late payment already spend an average of 86 hours a year on this.
Some of it needs you regardless: the client who rings to say they're in trouble, the disputed invoice, the decision to pause work on a long-standing name. Difficult payment conversations covers those.
The rest is bookkeeping about who gets what and when, and an agent can carry it inside the boundaries you set: which channels it may use, how persistent it's allowed to be, and what goes out without your sign-off. Adfin's Customer Agents work from your own email domain and pick the next action per client.
Two questions are worth keeping apart, though. Is there a best channel or send hour across all businesses and all clients? No reliable evidence settles that, ours included. Does this client reply to a text and ignore email? Their own record answers that, and an agent working from it doesn't need a general rule.
Common questions
How should you segment customers for credit control? By how they've paid you before. Work out the share of each client's invoices that arrived after the due date over their last five or more, then group them into never late, occasionally late, late most months, and late plus unresponsive.
Is it better to segment by invoice size or payment history? Payment history. In Adfin's book the median delay among late payments is 8 days in every invoice size band, and inside one business's own ledger 39 ran later on large invoices against 43 later on small ones. Big invoices need earlier confirmation instead of their own group.
How many chasing rules do you actually need? Four groups plus a default for new clients covers most ledgers, and five is about as many as anybody applies consistently by hand.
Where do you put a client with no payment history? On your default rules, usually your middle group. Two or three invoices tell you very little, and being firmer with new clients is hard to explain if they compare notes.
Should you write different reminder emails for different clients? Your schedule and your collection method are worth varying far more than the wording. Different tone for different clients reads as inconsistency, and the same message with a mandate behind it achieves more.
How often should you review your client groups? Quarterly. Any less often and a client who started slipping has three or four more invoices out before you notice, and monthly turns one bad month into a reclassification you then reverse.
Sources
- Small Business Commissioner — late payments research (accurate as of August 2026)
This article describes an approach to credit control and is not legal or financial advice. The concentration figures are measured on Adfin's own book, on repeat payers who made at least five payments themselves, so that population is selected and isn't a UK average. Last updated August 2026.
