Operated function · receivables
Your ageing report is not mostly bad debt. It is mostly invoices that went to the wrong person, arrived without a purchase-order number, got queued behind an approval nobody chased, or simply were never followed up because the person who would have done it had a harder week.
Sort it by reason rather than by age and the shape changes. A large share is administrative: the invoice went to a person who left, it lacked a reference the customer’s system requires, it was rejected by a portal nobody watches, or it sits behind an internal approval on the customer’s side that nobody has nudged.
Another share is people who are not going to pay soon and everyone quietly knows it. That group needs a decision, not a reminder — and the decision keeps getting deferred because making it means writing something off.
The rest is the group that would pay this week if the right person received a clear, unembarrassing request. That group is the money, and it is the group that gets worked least, because a polite follow-up on a small invoice feels like the lowest-value thing on anyone’s list.
So the work is uneven and it is personal. The person doing it knows which customers are sensitive, which contact actually pays, and which account you must not upset because a renewal is close. That knowledge is real and it is entirely undocumented, which means it does not scale and it leaves when they do.
And the reporting arrives monthly, as a total. A number that moves once a month cannot tell you whether last Tuesday’s effort worked.
Follow-up on an overdue invoice is nobody’s scheduled work, so it happens by available attention rather than by rule — and the invoices that get skipped are systematically the small and the awkward ones.
This is why hiring a collector helps less than expected and why an aggressive agency helps in the short term and costs in the long term. Neither addresses the actual mechanism, which is that the work is unscheduled. Add capacity to unscheduled work and you get more of the same unevenness at higher cost.
Making it scheduled changes the arithmetic. Every open invoice has a state, a next action, an owner, and a date by which that action should have happened. When that is true, the small invoice gets its follow-up because the follow-up costs almost nothing to send, and the awkward account gets its escalation because the escalation is a rule rather than a decision somebody has to feel brave about.
The second thing that changes is the reason code. Once follow-ups carry a recorded outcome — wrong contact, missing reference, portal rejection, internal approval, dispute, unwilling — the ageing report stops being a list of amounts and becomes a list of causes. Most of those causes are fixable upstream, permanently, in your invoicing rather than in your chasing.
What does not change is who decides. Whether to escalate an account, offer terms, pause a service or write something off is a judgement with commercial consequences, and it stays with your people under a written rule.
Days sales outstanding — measured by DSO against your own trailing baseline, with the mix held constant so a change in customer profile is not read as an improvement.
What the ageing is actually made of — measured by open balance grouped by recorded cause rather than by age — a breakdown most finance teams cannot currently produce.
Invoices that were never delivered or were rejected — measured by the count caught before due date, against the count previously discovered after sixty days.
Small overdue invoices that get worked at all — measured by follow-up coverage by invoice size band, before and after.
Consistency of escalation — measured by accounts escalated at the threshold your policy defines, as a share of accounts that crossed it.
Causes fixed upstream rather than chased repeatedly — measured by repeat occurrences of the same recorded cause on the same customer, quarter over quarter.
that unwilling accounts become willing, and no credit or write-off decision. Nothing here decides terms, pauses a service, escalates to legal, or reports to a credit bureau. Where an account genuinely will not pay, this identifies it earlier and with a documented history — it does not collect it.
Read access to your accounting or ERP system for invoice, payment and customer state, through documented interfaces. Nothing migrates and no second ledger is created — a parallel record of what a customer owes is a month-end reconciliation problem and it will be found at audit.
Where a customer uses an invoicing portal, that portal is monitored as part of the operation rather than treated as their problem. A rejected invoice nobody saw is the cheapest thing on this whole page to fix and the most commonly missed.
Writing anything back — a note, a status, a promise-to-pay date — is a separate permission from reading, granted per field, so a visibility exercise cannot quietly become a ledger change.
Every message goes out in your language, from your domain, under a script your finance and legal owners approved. Nothing improvises tone on an account, because a collections message that reads wrong costs more than the invoice.
Every contact carries a receipt: who was contacted, on what, when, with what result, under which authority. That record is the account history somebody needs when a customer disputes what was said.
Authority is narrow and enumerated. Sending an approved follow-up is delegated; agreeing terms, pausing a service, escalating to legal and writing off are not, and no threshold moves them across without a new grant.
Operational access is not permission to train. Your customer, payment and terms data does not become material improving anything serving another organisation — including a competitor who happens to sell to the same customers.
Legal will want the scripts and the escalation boundary, because collections language is regulated in some jurisdictions and reputationally live in all of them. Both exist as documents to review before anything is sent.
The account owners are the stakeholder most often skipped and the one most likely to object later. Bring them in at scoping: the rule they need to see is that a follow-up never contradicts an active commercial conversation, and that accounts near renewal route to them rather than to a cadence.
Where an obligation attaches through your contract or a jurisdiction, it is marked applicability-gated rather than presented as standing.
A single ageing band or customer segment — read-only, with no contact made — to produce the cause breakdown that the ageing report cannot currently show.
The observation phase sends nothing. It classifies what is open by cause: undelivered, rejected, missing reference, wrong contact, internal approval, disputed, unwilling. Most finance teams have never seen their own ageing in that shape, and it frequently reallocates the problem away from chasing entirely.
Some businesses should stop there. If the breakdown shows the dominant cause is an invoice field your own billing omits, the fix is upstream in billing and no operated follow-up is warranted — you would have bought a chasing service for a data problem.
If you continue, the first delegation is the pre-due courtesy reminder on one band: the lowest-risk message in the whole operation, on approved language, with escalation defined before anything is sent.
That is the right instinct and it is why the first phase contacts nobody — it classifies the ageing by cause and hands you the breakdown. If you do proceed, the first delegated message is the pre-due courtesy reminder, on language your own finance and legal owners wrote, from your domain, with accounts in an active commercial conversation excluded by rule. If a segment cannot tolerate that, it should not be in scope.
Then the observation phase will show that, and it will have cost you a read-only exercise to establish rather than an assumption. When the balance genuinely concentrates in a few unwilling accounts, this is the wrong purchase — those need a commercial decision and possibly counsel, and nothing here makes an unwilling payer willing. What it can still give you is a documented contact history for each one, which is worth having before that conversation.
Which is why the protected-account rule is established before scripts are written rather than after an incident. An account near renewal, in an escalation, or in an active negotiation routes to its owner instead of into a cadence, and that routing is a rule rather than a judgement made per message. If your book is entirely accounts like that, the honest scope is narrow — possibly the observation phase alone.
Automated reminders send on a schedule and record whether they were sent. The gap this addresses is the outcome: whether the invoice was delivered at all, whether a portal rejected it, which contact actually pays, and why a specific account has not. If your billing system can already produce your ageing grouped by cause, this is a duplicate and you should not buy it.
It would, and this is not that. The premise of the page is that most overdue balance is administrative, so the highest-yield action is a clear request to the right person rather than more pressure on the wrong one. Escalation stays a decision your credit owner makes, on your threshold, with the history in front of them. If the operation is measured on recovery alone it will eventually damage a relationship, which is why the cause breakdown and not the recovery number is the primary measure here.