There is a particular conversation that happens in every SME factory around the 25th of the month. The owner asks accounts what is outstanding. Accounts reads out a number. The owner names two customers he is annoyed about. Somebody is told to follow up. Nothing changes, because none of that was a decision.
The problem is that the outstanding figure is a result. It is the sum of a hundred small choices made over the previous three months, most of them on the shop floor and at the gate rather than in the accounts room. This article is about those choices — where the money actually sits, what it costs you, and the four places you can intervene before it becomes a phone call.
What late payment actually costs
Start with a number you can defend, because “customers pay late” is not a number.
Take a ₹6 crore engineering unit. Closing receivables are ₹1.15 crore. Terms on paper are Net 45.
DSO = receivables ÷ annual sales × 365
₹1,15,00,000 ÷ ₹6,00,00,000 × 365 = 70 days. Against Net 45 terms, that is 25 days of slippage.
A day of sales here is ₹6 crore ÷ 365 = ₹1,64,384. So 25 days of slippage is ₹41 lakh of your money sitting in other people’s businesses, permanently, on a rolling basis.
Funded on a cash credit limit at 12%, that is about ₹4.9 lakh a year in interest — and that is only the part you can see on a bank statement. The part you cannot see is the order you did not take in March because the CC limit was full of somebody else’s slow payment.
Every 10 days you take off DSO releases about ₹16.4 lakh of cash. Not borrowed — released. It is your own money arriving earlier, it costs no interest, and once you have it you keep it. Going from 70 days to 50 is ₹32.9 lakh. There is no other project in a factory this size that frees thirty lakh without buying anything.
That framing matters because receivables work is unglamorous and always loses the argument against a new machine. Put a rupee figure on twenty days and it stops losing.
Where the money actually sits
Now split that ₹1.15 crore by age. This takes twenty minutes and almost nobody does it.
| Bucket | Amount | Share | What it usually means |
|---|---|---|---|
| Not yet due (0–45 days) | ₹74,00,000 | 64% | Working as designed. Leave it alone. |
| 46–60 days | ₹19,00,000 | 17% | Slow, not stuck. Almost all recoverable by phone. |
| 61–90 days | ₹12,00,000 | 10% | Something is wrong — usually a dispute nobody logged. |
| 91–180 days | ₹7,00,000 | 6% | Needs escalation, not another reminder. |
| 180+ days | ₹3,00,000 | 3% | Provide against it. It is not really an asset. |
| Total | ₹1,15,00,000 | 100% | Past due: ₹41,00,000 |
Look at where the overdue money is. ₹31 lakh — three-quarters of everything past due — sits in the 46–90 day buckets, the ones everybody treats as ordinary slowness and nobody makes a fuss about. The 180-plus bucket, which absorbs most of the arguing, most of the emotional energy and most of the owner’s Sunday evening, holds ₹3 lakh. Seven per cent.
The oldest debt is the most irritating and the least valuable. It is also the least likely to be collected by effort, because if a year of asking has not worked, a fourteenth phone call will not either. Shifting one hour a day from the 180-plus fight to the 46–60 bucket is the highest-return move in the whole exercise.
None of which means writing off the old stuff quietly. Price it instead: a ₹1.5 lakh write-off at a 30% gross margin needs ₹5 lakh of new sales to replace. That arithmetic — not the irritation — is what should decide whether it goes to a lawyer, a settlement or a provision. If you are unsure what your real margin is, how to cost a job properly is the prerequisite, and the profit margin calculator will do the conversion.
A credit limit is two numbers, not one
Most factories set a limit in rupees and stop. ₹10 lakh for this customer, ₹3 lakh for that one, written down once when the account opened.
A rupee figure on its own is half a control, because it says nothing about how long the money is gone for. ₹10 lakh at Net 30 and ₹10 lakh at Net 90 are completely different businesses to be in. The first is a normal trading relationship. The second means you are financing a quarter of somebody else’s working capital, for free, while paying 12% for the privilege.
So record both, together, per customer — the amount and the terms. And review them when behaviour changes, not on an annual cycle. A customer who has moved from paying in 40 days to paying in 75 has told you something more useful than any credit report, and he told you months ago.
Sizing the limit
The honest test is not what the customer asks for or what his turnover is. It is: what could we survive losing? A single customer at ₹25 lakh of exposure inside a ₹6 crore business is not a credit decision, it is a bet on that customer’s solvency — and it deserves to be discussed as one, by the owner, not waved through by whoever answers the phone on dispatch day.
Concentration is worth checking at the same time. If three customers are 60% of your receivables, you do not have a credit policy, you have three relationships. That is not automatically wrong — plenty of good Tier-2 suppliers live that way, and running to an OEM’s schedule often means exactly this. But it should be a decision you have made, not a fact you discover during an ageing review.
The number the ledger does not show
Here is where most credit limits quietly stop working.
A dispatch is queued for a customer with a ₹10 lakh limit. Somebody checks the ledger. It shows ₹7.3 lakh outstanding — under the limit — so the truck loads.
| What is actually at risk with this customer | Amount | On the ledger? |
|---|---|---|
| Overdue invoices | ₹4,20,000 | Yes |
| Invoices raised, still within terms | ₹3,10,000 | Yes |
| Delivered, not yet invoiced | ₹1,40,000 | No |
| In production, material already bought | ₹6,80,000 | No |
| True exposure | ₹15,50,000 | — |
| Credit limit | ₹10,00,000 | Over by ₹5,50,000 |
The ledger was not wrong. It was answering a different question. It tells you what has been invoiced and not paid; it cannot tell you what has been committed and not invoiced. And for a manufacturer, the committed part is the dangerous part, because material bought to a customer’s drawing is not stock you can sell to anybody else — it is a special-purpose casting with his part number on it.
Factories that do check credit limits usually check them at invoicing — the one moment when it is too late to matter. The goods have gone. The only remaining lever is conversation.
Put the gate at dispatch
Which gives the whole article its one structural recommendation: the credit decision belongs at the delivery order, before the truck loads.
That is the last moment you hold something the customer wants. Five minutes earlier you have leverage and choices — part-dispatch, advance against the balance, hold pending a payment, escalate to the owner. Five minutes later you have an invoice and a hope.
In practice this means a dispatch cannot be approved by whoever happens to be at the gate. It needs an approval step that reads exposure, and a named person who can override it and is accountable for the override. Two rules make it work:
- Exceptions get logged, not argued. Overriding a limit is often correct — a long-standing customer, a genuine one-off, a payment already on the way. What is not correct is that nobody can later list which dispatches went out over the limit and who released them. A short weekly list of overrides changes behaviour faster than a policy document.
- Part-dispatch is the underused answer. The choice is rarely all-or-nothing. Sending 60% of the order against the available headroom keeps the line running, keeps the relationship intact and puts a concrete reason on the table for the payment conversation.
Their clock does not start when yours does
You raised the invoice on the 3rd. Net 45 means the 17th of next month. Except the invoice reached their accounts department on the 11th, and it went back once because your purchase order reference was missing, so it was booked on the 19th. Their system will pay 45 days from the 19th.
You have just given away sixteen days and you will record it as the customer paying late.
This is the least glamorous section of this article and probably the most profitable. Three habits close nearly all of it:
- Invoice on the day of dispatch, not at month end. Batching invoices into a month-end run is administratively tidy and can cost you two weeks on every invoice raised in the first half of the month.
- Put their reference on your document. Their PO number, their part codes, their delivery note number. An invoice that cannot be matched at their end does not enter the payment queue — it enters a tray.
- Get the GST details right the first time. A wrong GSTIN or place of supply is not a paperwork annoyance; it blocks the customer’s input tax credit, and no accounts department pays an invoice that costs them credit. Correcting it later means a credit note and a re-issue, and the clock starts again from zero. Our guide to GSTR-1 mistakes covers the ones that cause this, and the e-invoice walkthrough covers getting the IRN and QR right at source.
The daily call list
Reports tell you the state of things. Lists tell you what to do today. For collections you want a list, and it should be built on next-contact dates rather than purely on ageing — because the most valuable call in the whole process happens before anything is overdue at all.
“Just confirming invoice 2841 is in your system and approved for payment on the 17th.” Three minutes. You are not asking for money — you are confirming a process, which is a conversation anybody can have without awkwardness.
Short, factual, and it establishes that you notice. Most customers have a queue, and the queue is ordered partly by who is paying attention.
Accounts to accounts for the first two weeks. Then the sales owner, who has the relationship. Then owner to owner. What matters is that the ladder is written down before it is needed, so nobody has to decide in the moment whether this is the week to escalate.
Almost every Indian SME delay resolves into one of about six causes, and only one of them is unwillingness to pay: the invoice never arrived, the PO number was missing, the goods receipt was not booked at their end, an approval is sitting with somebody on leave, a rate or quantity is disputed, or the money genuinely is not there. Five of the six are process problems that a pre-due call finds and fixes while fixing is still cheap.
Log the reason, not the promise
“Will pay next week” is not information. It cannot be counted, compared or acted on, and it is what most follow-up notes contain.
“Invoice returned — our PO number not printed on it” is information. Written down twenty times, it stops being a collections problem and becomes a five-minute change to an invoice template. That is the difference between chasing money forever and fixing the thing that keeps producing the chase.
A short-supplied consignment, a rate difference, a lot rejected at their inspection. The customer very often does not tell you he is withholding payment — he simply does not pay, and for sixty days it looks exactly like slowness. A useful default: any invoice past due with no committed payment date is a dispute until proven otherwise, and should be routed to the sales owner rather than getting another reminder from accounts.
The lever most SMEs forget they have
Section 43B(h) of the Income-tax Act gets discussed in Indian factories almost entirely as a payables headache — the rule that says an amount payable to a supplier registered as Micro or Small is deductible only in the year it is actually paid, where payment falls outside the MSMED Act window (45 days with a written agreement, 15 days without). Our procurement guide covers that side.
Turn it around. If you are registered on Udyam as Micro or Small, that same rule applies to your customers when they pay you. Your payment terms stop being a commercial preference and become their tax exposure — which is a very different conversation to have on day 40, and one that tends to get returned rather more promptly than a reminder email.
It only works if you are actually registered and your customer knows it — so put the Udyam number on the invoice, not just in a file. And confirm the current position with your CA before you lean on it in a negotiation; the provision has been amended and interpreted repeatedly since it came in, and this article is not tax advice.
Six ways this goes wrong
1 · Chasing the oldest bucket instead of the biggest
Most of the collections energy in a factory goes into the debts that generate the most annoyance, which are almost always the smallest and least recoverable.
Cost: ₹31 lakh sitting in easy buckets while an hour a day goes to ₹3 lakh that a year of asking has not moved. Fix: age the ledger monthly and allocate effort by bucket size. Decide the old stuff by arithmetic — settle, litigate or provide — and stop re-deciding it weekly.2 · Checking the limit against the ledger
The ledger shows invoiced-and-unpaid. It cannot see goods delivered but not invoiced, or material already bought against an order in production.
Cost: dispatches approved at ₹7.3 lakh of visible exposure when the real figure is ₹15.5 lakh, against a ₹10 lakh limit. Fix: compute exposure as invoices + delivered-not-invoiced + committed production, and check that at the delivery order.3 · Invoicing at month end
Tidy for accounts, expensive for cash. An invoice for a 3rd-of-the-month dispatch that goes out on the 30th has lost 27 days before the customer’s terms even begin.
Cost: up to four weeks of DSO on half your invoices, entirely self-inflicted. Fix: invoice from the approved delivery order on the day of dispatch. It is the same work, done earlier.4 · No pre-due call
Follow-up starts when the invoice goes overdue, which is the exact point at which every cheap fix has expired and the only remaining option is asking for money.
Cost: process delays — five of the six common causes — discovered three weeks after they could have been fixed in three minutes. Fix: a next-contact date on every invoice, set about five days before due. It is one call, and it is the highest-yield habit in this article.5 · Recording promises instead of reasons
Follow-up notes full of “will pay next week” produce a diary, not data. Nothing can be counted, so nothing gets fixed upstream.
Cost: the same process fault regenerating overdue invoices indefinitely, because no one ever saw it repeat. Fix: log a reason code against every delay. When one reason appears twenty times, you have found a template or a workflow to change.6 · Sales is paid on the order, not on the money
The incentive lands when the order is booked or the truck leaves. From that moment the person with the relationship, the phone number and the leverage has no stake in whether the payment ever arrives — and collections falls to someone with none of the three.
Cost: the structural version of every other mistake on this list, and the reason they keep coming back. Fix: pay the incentive on collection rather than on dispatch. It costs nothing to implement and it realigns the one person who can actually move the needle.The five-step routine
Five buckets, twenty minutes. Allocate collections effort by where the money is, not by which customer is most irritating.
Recorded together on the customer master, reviewed when payment behaviour shifts rather than annually.
Invoices + delivered-not-invoiced + committed production. This is the number a limit is checked against.
Approval on the delivery order before loading, with a named approver and a weekly list of exceptions.
Next-contact dates rather than ageing alone. Log the reason for every delay, never just the promise.
Where this lives in OEMup
Every step above can be run on paper, and for one customer it should be. What a system is for is holding the policy so it applies to all four hundred invoices without anyone remembering, and putting the right number in front of the person at the gate.
The policy: limits and terms
Credit limits are set in system settings alongside payment and delivery terms and document numbering, so they are a company-level policy rather than a note in somebody’s diary. The Customer Master then holds the customer’s details, GST number, multiple addresses, contact persons and credit terms in one place — the amount and the term together, which is the point of the section above. Customer price lists carry agreed selling prices and minimum quantities per customer and auto-fill on every order, which quietly removes one of the six delay causes: the rate dispute that surfaces at payment time.
The gate: order to delivery order
A customer order is built as a quotation with tax, HSN and price auto-populated and its own payment terms, then converts to a delivery order. The delivery order tracks delivered against ordered and remaining quantities with dispatch location, vehicle, transporter and driver — and it carries an approval step. That step matters more than it looks: only an approved delivery order can be invoiced, which is exactly the dispatch gate this article argues for, and it is also what makes part-dispatch a first-class option rather than an improvisation.
The clock: invoicing same-day and complete
The GST tax invoice is raised from the approved delivery order, so invoicing on the day of dispatch is the default path rather than a discipline. Invoice type covers tax, credit, debit, retail, SEZ and export; IGST is calculated automatically; original, duplicate and triplicate PDFs and the e-way bill come out of the same step, with e-invoice IRN and QR generated directly. Getting the GST details right at source is what stops the credit note and re-issue that resets the customer’s clock to zero.
The follow-up: the daily call list
This is the screen the article is really about. OEMup groups three follow-up tools together: a daily call list driven by next-contact dates, a reusable terms library, and every unpaid invoice with overdue days and outstanding amount — the ageing view and the action list in the same place. Because the list is driven by next-contact dates rather than by ageing alone, the pre-due call is something the system can put in front of you, which is the difference between a good habit and a good intention.
Alongside it, Payments & Wallets tracks payments and holds advance amounts against a customer wallet — useful when the answer to an over-limit dispatch is an advance rather than a hold — and automatic payment reminders handle the routine chasing so the call list is reserved for conversations that need a person. Task management gives the escalation ladder somewhere to live, with due dates, recurrence and threaded comments, so “sales owner to call by Thursday” is a tracked item rather than a verbal instruction. And a custom dashboard can carry the DSO and ageing tiles with drill-through to the source invoice, so the monthly review starts from live figures instead of a spreadsheet somebody rebuilt on Sunday.
No ERP decides what a customer’s limit should be — that is a judgement about how much of his solvency you are willing to underwrite. It will not make the call, and it cannot see a dispute nobody logged. What it does is make the limit visible at the gate rather than after the truck has gone, put today’s calls on one screen, and keep a record honest enough that next month’s review is about decisions rather than recollections.
See your real ageing in 20 minutes
Book a demo and bring one month of sales invoices — we’ll load them, set a credit limit and terms on a customer, run an order through to an approved delivery order, and show you the call list and the overdue view the way your team would actually use them.
Book a Demo →The bottom line
Collections is the last place to fix a receivables problem and the only place most factories try. By day sixty the goods are gone, the paperwork is wherever it is, and the whole of your remaining leverage is the quality of a phone call.
The decisions that mattered were earlier and duller. What the limit was and whether it included the material already committed. Whether the delivery order needed an approval. Whether the invoice went out on the 3rd or the 30th, and whether it had their PO number on it. Whether anyone rang on day 40 to check it was in the system.
If you do one thing this week, age the ledger into five buckets and find out how much of your overdue money is in the 46–90 day range. On most shop floors it is around three-quarters — sitting in the buckets nobody argues about, recoverable by phone, and worth ₹16 lakh for every ten days you pull it forward.
Related reading: Manufacturing CRM in India for the enquiry-to-dispatch half of the same revenue cycle, Procurement Management for the payables side and the three-way match, The Hidden Costs of Manual Inventory for the other place your working capital is hiding, and How to Cost a Job for the margin figure that decides what a write-off really costs.
FAQ
What is DSO and how do I calculate it?
Days sales outstanding is the average number of days your money spends with customers: receivables ÷ annual sales × 365. A unit doing ₹6 crore with ₹1.15 crore outstanding is at 70 days. Compare it against your stated terms rather than against last year — if you sell on Net 45 and run at 70, the 25-day gap is the thing to manage.
How much is a high DSO actually costing me?
Multiply slippage days by daily sales. At ₹6 crore a year a day is about ₹1.64 lakh, so 25 days beyond terms is ₹41 lakh sitting with customers — roughly ₹4.9 lakh a year in interest on a 12% cash credit limit, before counting the orders you could not fund. Read the other way: every 10 days off DSO releases about ₹16.4 lakh of your own cash, once, and you keep it.
What should a credit limit be based on?
A limit in rupees is only half a control — record the amount and the terms together, because ₹10 lakh at Net 30 and ₹10 lakh at Net 90 are completely different exposures. Size it against what you could survive losing rather than what the customer asks for, and review it when payment behaviour changes rather than annually.
What is true exposure and why does the ledger not show it?
The ledger shows invoiced-and-unpaid. Real exposure also includes goods delivered but not yet invoiced, and orders in production against material already bought — which for a manufacturer is often the largest piece, because a casting with the customer’s part number on it cannot be sold to anyone else. A customer showing ₹7.3 lakh on the ledger against a ₹10 lakh limit can be carrying ₹15.5 lakh of true exposure.
When should I call a customer about payment?
About five days before the invoice falls due. Most Indian SME delays are process rather than refusal — the invoice never reached accounts, the PO number is missing, the goods receipt was not booked at their end, an approval is with somebody on leave. All fixable in a three-minute call beforehand, and nearly impossible afterwards, because by then you are asking for money instead of confirming a process.
Which ageing bucket should I chase first?
The biggest, which is almost never the oldest. In the worked ledger, ₹31 lakh — three-quarters of everything past due — sits in the 46–90 day buckets treated as routine slowness, while the 180-plus bucket that absorbs most of the arguing holds ₹3 lakh, or 7%.
Does the MSME 45-day payment rule help me get paid?
It can, if you are registered. Section 43B(h) allows a buyer to deduct an amount payable to a Micro or Small registered supplier only in the year it is actually paid, where payment falls outside the MSMED Act window (45 days with a written agreement, 15 without). Most manufacturers think of it as a payables problem; from the receivables side, being registered on Udyam turns your terms into your customer’s tax problem. Put the Udyam number on the invoice, and confirm the current position with your CA before relying on it.
Should sales incentives be paid on order booked or on cash collected?
On cash collected. If the incentive lands at order or dispatch, the person with the relationship and the leverage has no stake in whether the money arrives, and collections falls to someone with neither. Moving the trigger is usually the highest-yield receivables change a factory can make, and it costs nothing to implement.
Profit margin calculator · Break-even calculator · Inventory carrying cost · Inventory turnover · GST calculator — all free, no sign-up.