Automating dunning: reach the money in the bank sooner

An unpaid invoice is money the business has already laid out: for materials, wages, subcontractors. Until it is settled, the supplier is unwillingly financing the customer. How large that pre-financing is depends on the payment term - and that has risen. At 32.21 days (Creditreform) in the first half of 2026 it stood about 0.75 days (Creditreform) above the previous year and roughly two days above the 2023 figure (Creditreform). That sounds small, but it adds up across the entire book of open invoices: every extra day ties up capital that is missing elsewhere.

Small businesses face an additional squeeze. In the first half of 2026, large companies secured average payment terms of 35.51 days (Creditreform), while small firms received only 26.37 days (Creditreform) - a gap of around nine days (Creditreform). On late payment the picture reverses: small firms waited an average of 10.52 days (Creditreform) beyond the term, medium-sized ones 10.66 days (Creditreform), large companies only 6.71 days (Creditreform). Being small means granting shorter terms and still being paid later. Every day of receivables outstanding is tied-up cash, and mid-size firms carry that burden above average.

That this matters is clear from the losses. Debtors who cannot extend their overdraft with their bank use the open supplier invoice as short-term substitute financing; this effect regularly accounts for around half of all late B2B payments (Atradius). At the same time, corporate insolvencies reached their highest level in over a decade in 2025 (Creditreform) and continued to rise in 2026. Letting open items drift risks not only a late payment but, in the worst case, a total loss - often noticed only once the deadline has long passed. A process that reliably follows up every receivable is therefore less a matter of convenience than of risk management.

A proven ladder starts mild and only sharpens where needed. After the payment term has passed, a friendly reminder follows around day three to five, the first notice with a clear deadline around day ten to fourteen, the second notice stating the cost of late payment around day twenty to twenty-five, and after that a final deadline with a note about the next step. The exact spacing is a matter of taste and industry; what matters is that it is fixed in advance rather than renegotiated case by case.

Not every receivable deserves the same treatment. A small amount can be bundled or pursued at longer intervals, because otherwise the handling costs more than the loss. Long-standing customers with a clean payment history can get a gentler curve than new customers with no track record. And an amount above a set threshold deserves a tighter cadence. This differentiation is the reason a good rule is more than a calendar entry - it captures the risk logic of the business that previously existed only in the accounts clerk head.

Clean exceptions matter. Where a complaint is open, the run must not send a notice; such a case belongs on hold until the matter is settled. Where an instalment plan has been agreed, the run follows the plan rather than the original due date. And where a customer is already late but stays in contact, a personal call is often more effective than the next automatic stage. The run takes over the routine but does not replace judgement in the few cases that need it.

An automatic dunning run is only as good as the data it works on. It relies on unique invoice numbers, correct due dates and a clean customer master without duplicates. Just as important is that incoming payments are machine-readable and matched to the right item - via the electronic bank statement and automatic reconciliation. Where payments are matched by hand, exactly the gap opens up in which an already paid item keeps being chased.

For the two sides - invoice and payment - to meet, a reliable link between the invoicing and accounting systems is needed. Whether that runs through direct data integration or through maintained middleware depends on what is in place; what matters is that no duplicate data holding arises in which two systems keep different truths about the same item. Structured invoice data helps considerably here - anyone who has to implement the e-invoicing mandate from 2027 anyway is at the same time creating the clean basis for automatic reconciliation.

Finally, every stage has to be filed so it can be traced. For a later audit - and in a dispute before a court - what counts is proof of when which notice went out with what content. An orderly, audit-compliant filing of the documents, of the kind that arises when digitising documents, is therefore not an extra but part of the dunning process. The following list summarises what should be in place before the first automatic run.

Whether a dunning process works cannot be read from the number of notices sent - on the contrary, a lot of notices is rather a bad sign. The telling figure is receivables outstanding, days sales outstanding (DSO): the average time between invoice and incoming payment. It shows how long the money sits with the customer on average and makes every improvement visible. A running metrics report keeps this figure current instead of reconstructing it once a year from the annual accounts.

The goal is a DSO that sits as close as possible to the granted payment term - every day above it is delay that the run addresses. Even a reduction of a few days eases liquidity noticeably, because it affects the entire open book. Providers of automated solutions report up to 85 percent (Bilendo) less manual effort in the dunning run; regardless of the exact figure, the effect lies less in the working time saved than in the earlier arrival of cash. That there is room here is shown by the level of digitisation: in Germany, around 44 percent (EOS) of companies recently ran a receivables process that was only partly or barely digitised.

Alongside DSO, two further angles are worth watching: the share of overdue items in the total open book, and the average number of days in delay. If the overdue share rises while DSO stays stable, the problem concentrates in a few large items - a hint to tighten the cadence on high amounts. Where cash gets stuck in the order flow, a look into your own system data helps; how to make bottlenecks visible in it is described in the article reading processes from system data.