Collections & Recovery · By Lenext Team · Published on 2026-08-12 · 9 min read

Dunning journey: how to structure reminders that recover before day 10

How to design a dunning journey that covers 100% of the portfolio from day one — overdue bands, channels, settlement policy, compliance with consumer law and data protection, and the metrics that show whether it works.

More than 82% of B2B debts up to 10 days overdue are recovered. After day 20, the average rate falls to around 50% (B2B industry study, 2025; see also Serasa Experian on collections and recovery in 2026).

It is not a gentle curve. It is a cliff, and it happens in the first three weeks — precisely the period in which most operations are still organising the spreadsheet.

What makes that number uncomfortable is that it says nothing about the debtor. It says something about your operation.

The problem is not skill. It is headcount.

Ask a collections manager what percentage of the overdue portfolio gets any contact in the first ten days. The answer is almost always the same: the large ones.

That makes sense. A team of three or four people with hundreds of invoices falling due each week will prioritise face value. It is the rational decision given the available resource.

The side effect is that the tail of the portfolio — many small invoices adding up to a figure that is not small at all — ages without anyone having spoken to the customer. By the time someone finally calls, 25 days have passed and the conversation has changed in nature: it is no longer a reminder, it is a negotiation.

A dunning journey exists to solve that specific problem. It is not about chasing harder. It is about reaching everyone, at the right moment, without depending on who happened to have spare time that day.

What a dunning journey actually is

It is the translation of the collections policy into a sequence of automatic actions, triggered by time and calibrated by profile.

Three decisions define it:

1. When each contact happens (overdue bands, including the period before the due date).
2. Through which channel, and in what tone.
3. For whom — because a single journey for the whole portfolio is almost always the wrong journey for almost everyone.

What changes compared to traditional collections is not the message. It is coverage: the journey speaks to 100% of the portfolio from day one, and the human team moves to where automation cannot help — large, complex or already-negotiating cases.

A journey design that works in B2B

The table below is a starting point, not a decree. What matters is the logic: contact before the due date is the cheapest of all, and intensity rises with the delay.

Three observations that make a practical difference:

• The D-7 to D+3 block is what pays. It is cheap, almost entirely automatable, and resolves most delays, which are operational: lost invoice, goods under inspection, an approval stuck in the customer's own finance team.
• The journey has to stop by itself. A customer who has paid or negotiated must not receive the next message. It sounds obvious; it is the most common mistake and the one that damages relationships most.
• Recurring lateness deserves its own journey. A customer who is late every month is not forgetful: that is a behaviour, and it should feed back into their credit limit.

Segmentation is what separates a journey from spam

Tailoring the journey to behaviour is the strongest lever in recovery — and the most ignored. Three axes are enough to start.

A fourth axis matters for large portfolios: commercial weight. A customer representing 8% of revenue cannot enter the same automated track as an occasional buyer — not because they deserve less rigour, but because the conversation needs to involve sales before it escalates.

Settlement without a policy is expensive improvisation

This is the point most operations underestimate.

If the journey works, it will generate negotiations. And if there is no written rule for discounts, interest, penalties and instalments, every negotiator improvises according to that day's conversation.

It is exactly the same problem the policy solves at origination: subjective, unauditable decisions that are impossible to explain afterwards. The symmetry is worth saying out loud: the policy does not end at approval — it governs recovery too.

What has to be defined before the journey runs:

• Standard charges (late interest, penalty, indexation) and how they are calculated.
• Maximum discount per overdue band, and who can approve beyond it.
• Maximum number of instalments per value band.
• What happens to the credit limit during and after a settlement.
• How a broken settlement is handled — because it will happen, and the second agreement cannot be more generous than the first.

Collecting while preserving evidence

In 2026, a good dunning journey collects and preserves evidence. Two layers of requirements apply at the same time.

Consumer protection law — in Brazil, articles 42 and 42-A of the Consumer Protection Code are the reference even in B2B operations, because they set the standard of conduct courts tend to apply by analogy: the debtor must not be exposed to ridicule, embarrassment or threat. Established good practice includes contacting during business hours (typically 8am to 8pm), avoiding weekends and public holidays, limiting attempts per day and per week, not disclosing the debt to third parties (other than guarantors), and clearly stating amount, origin and creditor.

Data protection (LGPD) — processing personal data for collections must respect purpose, necessity and transparency. In practice: tell the customer, in the contract or in a specific notice, that their data will be used for collections; do not repurpose data collected for something else; and keep a documented legal basis for data enrichment and cleansing, which is where compliance asks first.

And there is a layer that is not law but decides litigation: traceability. Every message sent, every contact logged, every offer made and refused has to be recorded. Many operations stall the scaling of collections precisely here — automation advances, the audit trail does not follow, and legal starts acting as a brake.

Stale data collects from nobody

A meaningful share of the overdue portfolio is not resistant. It is unreachable: disconnected phone number, the email of an employee who left, an address from three moves ago.

Data cleansing and enrichment fix that — and change the economics of outsourcing. One of the historical reasons for sending cases to an agency is that the agency has skip-tracing tools and the company does not. Having that capability in house means recovering without paying a percentage of what is recovered, and sending out only what genuinely has to leave — already with valid data.

Agencies: the inventory nobody can report on

When a case goes to an agency, it usually leaves the system too. Custody moves into an email inbox.

Run the test in your own operation: ask right now how many cases sit with each agency, how many days each one has been stalled, and which pending item is blocking resolution. The time it takes to answer is the diagnosis.

An unanswered email becomes a stalled case. A stalled case becomes a loss — without appearing in any report. The minimum to demand from a mature operation: a consolidated view of inventory by agency and stage, ageing of externally held cases, and performance comparison between agencies. Knowing which one recovers more, on comparable portfolios, changes the next contract renewal.

What Pix Automático changed

Since Brazil's regulatory cut-off in January 2026, companies that used traditional direct debit for recurring billing had to move to Pix Automático, and adoption accelerated through the first half of the year.

For anyone collecting, the practical effect is less friction in recurring billing: no expired cards, no declined authorisations, none of the involuntary default that has nothing to do with ability to pay. And unlike traditional direct debit, it does not require a complex banking arrangement — which opens recurring billing to operations that previously had no access to it.

It does not replace a dunning journey. It removes an entire class of delays the journey should never have had to handle.

The metrics that say whether it works

Compare DSO against your own payment-terms policy, not against a market figure. If the policy is 30 days and the average is 32, that is normal. If it is 60, there is a problem — and it may sit in origination, not in collections.

Collections is not the department that chases losses. It is the last stage of a credit decision made much earlier — and the only one you can still correct while the money exists.

What percentage of your overdue portfolio gets any contact in the first ten days?