Days Sales Outstanding · Part 4
How should AR answer a DSO question?
October 31, 2026
Three parts in, DSO has taken a bit of a beating. This is not a call to bin it. DSO is what gets forecast. Nobody forecasts ageing buckets. It sits in the board pack, it can be benchmarked, and as Part 1 said, it is a legitimate working capital ratio. It will be on the slide next month.
So if DSO stays, how should AR answer for it? This part follows a DSO question to AR’s desk and builds the answer step by step.
When the question is fair
In this series it only lands there after two checks. Part 2 asks whether the movement is a signal. Part 3 asks whether it sits in receivables or sales.
Shift C passed both. Eight months running about a day above the baseline, sales flat, and all 0.9 days in receivables. So AR gets the question: what has kept receivables higher since February?
Days into money
You can’t chase a day. Money sits with customers and invoices, and that is where the answer is.
0.9 days × 12.0m / 30 is 0.36m. That is the amount AR needs to find.
Where it sits
The first split is between money not yet due and money overdue. A rise in balances not yet due usually points to terms, customer mix or a large order. Customers don’t owe that money yet. A rise in overdue balances is money customers should already have paid.
| Baseline year | Feb to Sep 2026 | Change | |
|---|---|---|---|
| Not yet due | 15.20m | 15.20m | 0.00m |
| 1–30 days overdue | 2.40m | 2.48m | +0.08m |
| 31–60 days overdue | 0.80m | 0.92m | +0.12m |
| Over 60 days overdue | 0.44m | 0.60m | +0.16m |
| Total | 18.84m | 19.20m | +0.36m |
Averages of the month-end ageing in each period. Sales, terms and customer mix are held flat in this scenario. Made-up numbers.
Nothing moved in balances not yet due. The whole 0.36m is overdue, and the biggest increase is in the oldest bucket. Overdue is up 10%. Over 60 days is up by more than a third. If the monthly ageing shows the rise starting in February, it matches the shift in Part 2.
How many accounts carry it
Next, sort the overdue increase by customer, largest first. If a handful of accounts carry most of it, the answer is about them: a dispute, a customer in trouble, a payment run that slipped. The question then is what happened to those customers and how much risk they now carry.
If the largest increases add up to only a small part of the gap, say 0.05m of the 0.36m, the rest is spread thinly across many accounts. There is no customer to point at, and that is the finding. When an increase is that broad, something changed in how the book is worked, not in any one customer.
A broad increase narrows the search to things that hit many accounts at once. Two are worth ruling out before AR takes ownership. Unapplied cash: payments that arrived but were never allocated leave invoices showing as overdue, and if the unapplied balance is netted against balances not yet due, the total looks right while overdue looks worse. Billing: an invoicing error or system change can produce small disputes and short payments everywhere.
If both are flat, what is left is how the book is worked, and what changed in February. A change in dunning strategy. A team that has changed shape, or been given different objectives. Or simply a team that is short. Any of them shows up the same way: a little less attention on every account, and the oldest debt chased last.
No single month broke the limits. It didn’t need to. A drift that keeps going one way, however small, deserves a question once it is clear something isn’t right. That is not a question for every month. For some it is obvious after three months, for others it takes longer.
Where to look first
Where it sits and how many accounts carry it give four combinations. Each one says where to look first, not what the answer is.
| A few accounts | Many accounts | |
|---|---|---|
| Not yet due | Specific contracts or terms, a large order | Standard terms changed, a shift in customer mix |
| Overdue | What happened to those customers: a dispute, financial trouble, a missed payment run | Something in how the book is worked: capacity, process, reminders |
The top row usually goes back to Sales or billing. The bottom left is AR’s to chase, account by account. Only the bottom right points at how the department is running. Shift C sits there.
What happens to the baseline
Part 2 said to recalculate the baseline after known changes. That holds for changes that are deliberate and here to stay. New standard terms move DSO to a new level, it stays there, and the limits should follow.
Shift C is not one of those. It is a problem AR intends to fix, so the baseline stays where it is, and so does the forecast. If the fix works, the chart will show DSO coming back down. Reset the baseline now and the chart stops showing the problem you are trying to fix.
Checklist: is the answer ready?
- Is the movement stated in money, not days?
- Does it sit in balances not yet due or overdue, and in a few accounts or many?
- If many, have unapplied cash and billing been ruled out?
- Does each part have an owner, and does AR’s part have an action?
- Is the change here to stay, or one to fix? Does the baseline move?
- What does the evidence show, and what still needs explaining?
From variance to performance
Back to Part 1’s two questions. The variance question was why DSO came in higher than expected. In shift C, AR has answered it: more overdue, spread across the book, coming from how the book is worked.
The performance question was whether the collections team did a worse job. Here the ageing is blunt. Overdue growing across the book, fastest in the oldest bucket, is AR’s to fix, whatever sits behind it. A short team explains it. It doesn’t excuse it. The same goes for a dispute left past 60 days because someone else has to act. Overdue is a red flag from the first day, not from the next review.
DSO needed a control chart and a split to get here. The ageing showed it directly. If not DSO, then what? Part 5 looks at what to measure instead.