Automate the dunning process: a practical guide for SMEs
An invoice is written, the work is done – and then: nothing. The money does not arrive, the reminder gets lost in daily business, and only weeks later does anyone notice that a sizeable amount is still outstanding. When you automate the dunning process, you take exactly that forgetting out of the picture: every due invoice is detected, chased politely and on time, escalated in stages if needed and tracked through to payment. For small and mid-sized companies this is not a dull corner of accounting but a direct lever on cash flow – because money that comes in on time does not have to be bridged with an overdraft.
This guide is aimed at owners and managers as well as those responsible for accounting and finance who want to collect receivables more reliably without typing every reminder by hand and without straining good customer relationships. It explains why overdue invoices so often get left, what automation means here in concrete terms, which building blocks and tools work together, where an AI agent helps – and what a realistic start looks like in five steps. With a fictional example from an SME that shows how the work shifts from searching and chasing to calm monitoring.
In brief
- No due date missed: The system detects overdue invoices automatically and starts the chase, instead of waiting for someone to scan the open-items list.
- Friendly first, firm later: A polite payment reminder before the first formal notice saves many customer relationships – the staged flow stays binding nonetheless.
- AI in the right place: An AI agent reconciles payments, drafts suitable letters and suggests a reply to queries – a human decides.
- A workflow as the hub: A tool such as n8n ties accounting, bank and email together – even across many third-party systems.
- Make cash flow visible: Every invoice feeds data on payment behaviour. You can see which customers pay late – and steer accordingly.
Why overdue invoices get left
Chasing payments is one of the least popular tasks in any company: it is uncomfortable, it takes time, and it earns no money directly. That is precisely why it gets put off. The problem is rarely ill will but a missing process: the open items sit in the accounting software, the payments on the bank account, and who follows up depends on whether someone happens to have time to reconcile the two. If a colleague is off or it is peak season, a due invoice can easily sit unnoticed for three weeks – and a short reminder turns into a real bad debt.
On top of that comes the worry of annoying a good customer with a reminder. So you wait "a few more days", and the days become weeks. For small and mid-sized companies this delay is expensive, because the missing money is missing elsewhere – on the next material order, on wages, on your own invoice from a supplier. Without structured capture there is also no analysis: which customers consistently pay late, and at what amounts is earlier chasing worthwhile? You can only see that if due dates and payments come together. This is exactly where the idea of automating the dunning process comes in, rather than pursuing every overdue invoice by hand as a one-off.
What "automate the dunning process" really means
"Automatic" does not mean a machine fires off cold dunning letters at random. It means the system takes over the repetitive, error-prone steps – detecting due invoices, reconciling them against payments received, sending a reminder on time and escalating to the next stage when a deadline passes. The human stays where it matters: handling regular customers, disputed invoices and the decision of when a case goes to a collection agency or a lawyer.
The degree of automation can be finely dosed. At the simple end there is an automatic payment reminder that gently nudges a few days after the due date. One step further, a workflow checks the open items daily, reconciles them with the bank transactions and starts a staged dunning run with fixed deadlines for every overdue invoice. At the top end, an AI agent drafts individual letters, recognises partial payments and prepares a reply to queries. What matters is that each stage delivers value on its own – even the automatic first reminder brings in a large share of payments without anyone having to remember. If you want to cover the invoicing side too, our article on automating accounting in a small business provides the frame.
A practical example from an SME
Fictional example, not a customer case: A wholesaler for catering supplies with fourteen employees issues around 200 invoices a month with a 14-day payment term. Until now, accounting looks at the open-items list "when there's time", compares it by hand with the bank statements and types individual reminders. Some invoices are chased after a week, others only after two months, depending on workload. There is no overview of which customers generally pay late.
In the automated process, a workflow checks the open items every morning and reconciles them with the bank transactions. If an invoice is three days past its due date, a friendly payment reminder goes out automatically by email – with invoice number, amount and a payment link. If payment stays out, first and second dunning stages follow after fixed deadlines, each with a polite-but-firm text. Before every escalation, accounting reviews the suggestions and can hold back regular customers with a click. Disputed invoices and anything past the final stage land as a task on the managing director's desk. "Whoever has time" becomes "every due invoice is chased on time and consistently". How much this speeds up payments depends on the business; plug in your own numbers before you judge the benefit.
From due to paid: the building blocks
An automated dunning process consists of a few clearly delineated parts. Set them up cleanly, one after another, and you get a result that works reliably and can be extended later:
- Detect open items: The flow regularly reads from accounting which invoices are due or overdue – instead of someone scanning the list by hand.
- Payment reconciliation: Bank transactions are matched to open invoices so that no one who has already paid gets chased – the most common and most embarrassing mistake in dunning.
- A reminder before the notice: A few days after the due date a friendly reminder goes out – often that is enough, because the invoice had simply been overlooked.
- Staged dunning with deadlines: If payment stays out, first and second notices follow with clear deadlines, suitable texts and, optionally, late fees and interest.
- Approval and exceptions: Regular customers, partial payments and disputed cases can be held back; the final stage before collection or a lawyer is always decided by a human.
- Analysis: All cases feed into a simple statistic that reveals payment behaviour per customer and provides the basis for better payment terms.
The most important and most frequently underestimated building block is clean payment reconciliation. As long as it is not reliably clear which invoice has already been paid, you send reminders to the wrong people and lose exactly the trust you want to protect. It pays to set up the match between accounting and bank account reliably first – everything else builds on that. The topic is related to the e-invoicing mandate 2027, because structured invoice data makes automatic reconciliation much easier.
Where the AI agent makes the difference
Artificial intelligence lifts the dunning process where language and unclear cases come into play. The tone of a payment reminder helps decide whether a customer pays and stays or feels hounded. An AI agent can draft a suitable letter depending on customer history and dunning stage – friendly for the first reminder, firmer for the second notice – which accounting only needs to review and approve. If a customer replies with a query or announces a partial payment, the AI reads the message, assigns it to the right invoice and suggests the next response. How to use such agents responsibly is set out in our overview of AI agents.
Two limits matter here. First, the AI makes no final decision on deferral, waiver or handing a case to a collection agency – it prepares, the human decides and owns the relationship with the customer. Second, payment and customer data is sensitive information: make sure processing is GDPR-compliant and runs, as far as possible, on European servers or in-house. Used correctly, the AI agent speeds up the groundwork and ensures a consistent, fair tone without shifting responsibility.
Which tools fit: n8n and your accounting
For the technical implementation, your setup matters most. Often your data sits in many different systems – your own accounting software, a shop, a separate bank tool. This is where n8n plays to its strengths: it can be self-hosted, connects to almost any interface and makes every step visible and auditable – an advantage precisely with sensitive payment data. The decisive factor is the connection to your accounting and bank: without reliable reconciliation, any dunning run stays patchy. If you would rather not build the process yourself, our dunning and payment-reminder automation offers a ready starting point. What matters is not the logo but that the flow fits cleanly into your existing landscape.
Five steps to an automated dunning process
- Write down the current process: Note down honestly how an overdue invoice is pursued today – from due date to payment. The biggest gaps and waiting times quickly become visible.
- Define stages and deadlines: Set a lean ladder – say reminder, first and second notice – with fixed deadlines, texts and the question of when late fees or interest apply.
- Set up reconciliation: Make sure payments are reliably matched to invoices. This step looks unspectacular but decides whether you chase the right people.
- Connect the workflow: Tie accounting, bank and email together with a workflow tool such as n8n into a process that detects due invoices, reminds and escalates in stages – and, if useful, add an AI agent for texts and queries.
- Test, analyse, hand over: Run the process on real open items, check texts and exceptions and clarify ownership and upkeep before it becomes the standard.
Legal context, common mistakes and limits
A brief look at the legal context belongs here: in Germany a business customer is generally in default at the latest 30 days after the invoice is due and received, and even the first notice is often not legally required – a polite payment reminder is good practice nonetheless and preserves the relationship. From default onwards, interest and flat-rate collection costs can be claimed. Exactly how you design stages, deadlines and fees should, in case of doubt, be agreed with your tax adviser and lawyer: this article is a practical orientation and does not replace legal advice.
The most common mistake in implementation is skipping reconciliation and starting straight with sending – chasing invoices that have already been paid loses customers faster than any acceleration gains. Equally common is too harsh a tone from the start: the first message should always be a friendly reminder, not a threat. And finally, automation must not mean that no one looks any more – regular customers, partial payments and disputed invoices still need a human to decide with judgement. Plan for that and you end up with a dunning process that strengthens cash flow without sacrificing the customer relationship – and that turns an unpopular duty into a calm, reliable routine.
Want your overdue invoices collected faster and more reliably?
Lyron sets up a dunning process with you that fits your systems – from payment reconciliation through reminders and staged notices to the approval of exceptions, as a workflow in n8n, optionally with an AI agent for letters and queries. We start with your standard cases and hand over a process your team can maintain itself.
Free Consultation