Home / Blog / Predicting Late Payers

Collection

How Do You Predict Which of Your Customers Will Pay Late?

Late payers rarely surprise you if you watch the signals. Learn which payment-history and behavior signals to track, how to score risk and what to do early.

October 20, 2026 · 5 min read

Late payment rarely comes out of nowhere. Most customers who pay 40 days late this quarter were already drifting last quarter, and the signs were sitting in your ledger. You do not need a data-science team to spot them. With a handful of signals and a simple score, you can tell which accounts deserve an earlier reminder, a tighter credit limit or a phone call before the invoice is even due. This guide covers the signals worth tracking, how to combine them into a risk score, and what to do once an account turns amber or red.

Start with payment history

The best predictor of how a customer will pay is how they have paid. Look at the last six to twelve invoices for each account and track:

• Average days to pay. Compare it with your terms, not with other customers.
• Share of invoices paid late. A customer late on one invoice in ten is a very different risk from one late on six in ten.
• Broken promises. A promised payment date that passes without payment is a stronger warning than lateness alone.
• Short payments and disputes. Partial payments and repeated disputes often come before slow payment.
• Bounced or reversed payments. One is noise; two is a pattern.

Watch the trend in days beyond terms

A single average hides the story. A customer who averages 12 days late may be stable at 12, or may be climbing from 3 to 21. For every invoice, record the days beyond terms, meaning the gap between the due date and the day the money arrived, and look at the direction over the last three to five invoices.

A rising trend matters more than a high but steady number. A customer who has paid exactly 10 days late for two years is predictable, and you can plan your cash around it. A customer going 2, 6, 11, 17 days late is telling you something has changed: cash pressure, a new approver, or a quiet dispute. That signal often arrives weeks before the first seriously overdue invoice.

Add order and behavior signals

Payment history looks backward. Order and behavior data can show trouble earlier.

• A sudden jump in order size. A large order far above normal, especially from a new customer, raises your exposure fast.
• Unusual ordering patterns. Rush orders, orders split to stay under a limit, or new orders placed while an old invoice is still unpaid.
• Slower communication. Unanswered emails, new contacts appearing, or repeated requests to resend the invoice.
• Credit utilization. An account sitting near its limit has little room to absorb any delay of its own.
• External signals. News of layoffs or restructuring, or a drop in a credit-bureau score, where you have access to one.
• Limited history. A new customer has no record yet, so treat it as medium risk until it builds one.

Build a simple risk score

You do not need a statistical model. A points table that anyone on the team can read is easy to explain and easy to adjust. The weights below are illustrative, so tune them against your own data.

• Average days beyond terms over 15: 2 points; over 30: 3 points
• Worsening trend across the last three invoices: 2 points
• More than 30% of invoices paid late: 2 points
• Broken promise to pay in the last 90 days: 2 points
• Open dispute or short payment: 1 point
• Balance above 80% of credit limit: 1 point
• New customer with fewer than three invoices: 1 point

Add up the points. A score of 0 to 2 is low risk, 3 to 5 is medium, and 6 or more is high. Review scores monthly and then check them against reality: of the accounts you flagged as high risk, how many actually paid late? If the score is wrong, change the weights. A transparent, rough score that people use beats a clever one that nobody trusts.

What to do for high-risk accounts

A score is only useful if it changes what you do. For high-risk accounts, consider:

• Remind earlier. Send the courtesy reminder 7 days before the due date instead of 3, and follow with a call on the due date.
• Confirm the invoice is approved. Ask who signs it off, whether it was received and whether it matches the purchase order.
• Tighten credit. Lower the limit, ask for a deposit or partial prepayment, or hold new orders once the account passes a set number of days overdue.
• Shorten terms or add an incentive. Net 15 instead of net 30, or a small discount for early payment.
• Make paying easy. A payment link, direct debit or card on file removes excuses.
• Name an owner. One person should be responsible for the account and agree a payment plan early if needed.

Medium-risk accounts get a lighter touch: an earlier reminder and a note on the account. Low-risk accounts stay on your standard sequence, so you spend effort where it actually pays off.

Common mistakes to avoid

• Treating every late payer the same. A customer 5 days late once and one slipping steadily need different responses.
• Reacting only after the due date. By then you have lost your best weeks to act.
• Punishing good customers for one slip. A single late payment after years of reliability is usually noise, so look at the pattern first.
• Setting credit limits once and forgetting them. Limits should follow the score, not the day the account was opened.
• Never checking the score against outcomes. Without feedback, the weights drift away from reality.

Start with three or four signals you can already pull from your ledger, score your accounts this month, and refine from what you see.

Frequently asked questions

Can you really predict which customers will pay late?

Not with certainty, but you can rank accounts by likelihood. Payment history, the trend in days beyond terms and a few order signals are usually enough to separate low-risk from high-risk customers.

What is the best single signal of late payment?

The trend in days beyond terms. A customer paying a little later each time is a stronger warning than one who is steadily late by the same amount.

How many invoices do I need to score a customer?

Six to twelve invoices give a reliable picture. For newer customers with fewer than three, treat them as medium risk until they build a history.

What should I do first for a high-risk customer?

Send the reminder earlier, confirm the invoice is approved and received, and consider a lower credit limit or a deposit before taking on large new orders.

← Back to the blog