Every CFO under working-capital pressure receives the same instruction, and it is almost useless: reduce DSO.
Days sales outstanding is a symptom measure. It compresses four structurally different problems into one number, which is why two organisations with identical DSO can require entirely different interventions — and why a collections programme that transformed one business releases nothing in another.
Worse, two of the four traps are not finance problems. Applying collections pressure to them achieves nothing except a slower deterioration in customer relationships, because the reason the customer has not paid is that finance sent an invoice they cannot pay.
The governing rule
Cash is trapped at the point where the obligation became unclear — not at the point where the chasing happens. Collections is the last step in the cycle and therefore the place where every upstream defect eventually appears. Treating the place where a problem appears as the place where it originates is the central diagnostic error in working capital.
The four traps
Trap 1 — The invoice was never collectable
The invoice was late, wrong, sent to the wrong entity, missing a purchase-order reference the customer's AP system requires, or non-compliant with a statutory e-invoicing format. It was never going to be paid on terms, because it could not enter the customer's payment run.
This is not a collections problem. It originates in pricing, contracting, fulfilment or billing configuration. Collections effort applied here produces a specific and recognisable pattern: high contact volume, high dispute rate, low conversion, and an account manager increasingly reluctant to let finance near the customer.
The test. Take your aged debt beyond terms and ask what proportion is disputed, queried or awaiting a corrected invoice. If it is above roughly a fifth, this is your primary trap and no collections tooling will fix it.
Trap 2 — The cash arrived but was not applied
Payment is in the bank and the invoice is still showing open, because the remittance could not be matched — aggregated payments, malformed references, short payments with no explanation, remittances arriving as PDFs in a shared mailbox.
Economically this is the strangest trap: the cash is already yours. What is trapped is the information, and the consequences are real — customers chased for invoices they have paid, credit limits blocking orders unnecessarily, and a receivables ledger nobody trusts.
The test. What proportion of receipts are applied automatically on the day they arrive? Below roughly two-thirds, cash application is your constraint. This is the most automatable step in the cycle and the one where AI-assisted matching has the clearest genuine value.
Trap 3 — Nobody decided who to chase
The invoices are correct, the cash is applied, and collections is being performed — but without segmentation. Every overdue account receives the same treatment, so scarce collections effort is spread evenly across accounts that would have paid anyway and accounts that will never pay.
This is the trap that collections platforms are actually built for, and where prioritisation by risk, value and payment behaviour produces the fastest measurable improvement.
The test. Ask your collections team how today's worklist was produced. If the answer describes an ageing report sorted by value, or individual judgement, you have Trap 3 — the work is being done, but not directed.
Trap 4 — The terms were wrong before anyone shipped
The customer is paying exactly as agreed, and what was agreed is uncommercial: terms conceded in a negotiation nobody in finance saw, or credit extended to a customer whose risk was never assessed.
No collections activity can recover this, because there is nothing overdue. It is recovered only at the next contract, and prevented only by putting credit and terms policy upstream of the sale.
The test. Compare weighted average agreed terms against your DSO. If DSO is close to agreed terms but both are well above your peer set, your collections function is working correctly on a commercially poor book.
| Trap | Where it originates | Diagnostic signal | The lever that works | The lever that does not |
|---|---|---|---|---|
| 1 · Invoice never collectable | Pricing, contracting, fulfilment, billing config | >~20% of overdue debt disputed or queried | Invoice accuracy and e-invoicing compliance at source | Collections effort — chases a defect faster |
| 2 · Cash arrived, not applied | Remittance capture and matching | <~66% of receipts auto-applied on day of receipt | Cash application automation, remittance capture | Collections effort — chases invoices already paid |
| 3 · Nobody decided who to chase | Collections operating model | Worklist is an ageing report sorted by value | Segmentation and prioritisation by risk, value and payment behaviour | More collections headcount applied evenly |
| 4 · Terms were wrong | Sales negotiation, credit policy | DSO ≈ agreed terms, both above peer set | Credit policy and terms governance upstream of the sale | Anything downstream — nothing is overdue |
Why the order matters
The dependency ladder in the order-to-cash cycle is not a preference. Electronic invoicing precedes cash application, which precedes collections — because you cannot match a payment against an invoice that was never correctly issued, and you cannot sensibly chase an invoice that has already been paid but not applied.
This is why collections automation so often disappoints. It is frequently the first investment, because collections is where the pain is felt and where the software demonstrates most vividly. If Trap 1 or Trap 2 is the real constraint, a collections platform simply industrialises the chasing of defects.
| Order | Move | Releases | Typically visible in |
|---|---|---|---|
| 1 | Fix invoice accuracy, delivery and format compliance at source | Removes the disputed-debt population that no chasing recovers | 1–2 quarters |
| 2 | Automate remittance capture and cash application | Converts already-received cash into visible, applied cash | 1–2 quarters |
| 3 | Segment and prioritise collections by risk, value and payment behaviour | The genuine DSO reduction | 2–3 quarters |
| 4 | Move credit and terms policy upstream of the sale | Prevents the next book being built the same way | Next contract cycle |
Steps 1 and 2 are also, conveniently, the ones that release cash fastest and require least technology — which is what makes working capital the natural first funder of a wider transformation, as set out in our business case framework.
Where technology is warranted, our Accounts Receivable Buyer's Guide and the 2026 ranking grade the platforms against exactly these capabilities — collections, cash application, credit management and electronic invoicing — and are explicit that the leading platforms are not equally strong across all four. Which one leads depends on which trap you have.
The relationship cost nobody models
Receivables is the only finance process with a direct line to the customer. Every automation decision in this cycle is also a relationship decision, and business cases almost never model that side.
Automated dunning applied without segmentation will, over a year, contact your best customers about small balances with the same insistence it applies to genuine delinquency. The cash released is measurable. The commercial goodwill spent is not, which is precisely why it gets spent.
The practical discipline: segment before automating, exempt strategic accounts from automated sequences by policy rather than by exception, and give the account owner visibility of what finance is about to send. None of this is expensive. All of it is routinely skipped.
The strongest case against this diagnostic is that the traps are not cleanly separable in practice. Real receivables books have all four at once, in proportions that shift by customer segment and geography. A CFO could spend a quarter diagnosing and still not have a single clear answer — while a straightforward collections push would have produced visible cash in the same period.
That is a fair challenge, and it is why the diagnostic tests are deliberately crude. Each uses data a finance function already holds and can be run in an afternoon, not a quarter. The purpose is not to partition the book precisely; it is to establish whether the dominant constraint sits upstream or downstream of collections — because that single distinction changes which investment makes sense.
There is a second objection worth conceding. Our DSO benchmark is a peer-set model, and DSO varies enormously by industry, contract structure and customer mix. An organisation selling to public-sector customers on ninety-day terms is not underperforming because its DSO exceeds a cross-industry median. Where the peer comparison is weak, Trap 4's test — DSO against your own weighted agreed terms — is the more reliable one, because it is internal.
What to do with this
Run the four tests. They need one aged debt report, one cash application statistic, one conversation with the collections team and one comparison of DSO against agreed terms. Most finance functions can complete this inside a week.
The output is not a number. It is a sentence: our cash is primarily trapped at trap N. That sentence determines whether the next investment is billing accuracy, cash application, collections prioritisation or credit governance — four different programmes, four different sponsors, and only one of them correct.
- DSO is a symptom, not a diagnosis. Four structurally different traps produce the same
headline number, and only one of them responds to chasing harder.
- Two of the four traps are not finance problems. Uncollectable invoices and uncommercial
terms originate in billing, contracting and sales; collections pressure applied there spends customer goodwill and releases nothing.
- Fix upstream first — it is faster and cheaper. Invoice accuracy and cash application
release cash within one to two quarters with little technology, and they fund the collections capability that follows.