Almost every small business has money sitting in someone else's account that should be in theirs. The instinctive explanations — customers are struggling, big firms squeeze small ones, it's the economy — are sometimes true. But in most books of overdue invoices, the largest single category is duller than that: nobody asked, or nobody asked twice.
Chasing is a job that loses every scheduling contest. It's uncomfortable, it's nobody's actual role in a small team, and the person who'd have to do it is the same person delivering the work. So it gets done in bursts when cash gets tight — which is the worst possible timing, because by then the invoices are old, the buyer has moved on, and the conversation is loaded.
Why timing beats tone
The single biggest lever is not how firmly you ask, it's how soon. An invoice queried at day 32 is an admin matter. The same invoice at day 90 is a negotiation, often about work the buyer has half forgotten, sometimes with a claim attached that would never have appeared if you'd asked in week five.
Early chasing also surfaces the real reasons, which are rarely refusal to pay:
- The invoice went to the wrong person, or into a portal nobody monitors
- A purchase order number is missing, so it's sitting in a queue unapproved
- It's genuinely in the next payment run and everyone can relax
- There's a query about one line, and the whole invoice is parked because of it
- They've hit a cash problem and would tell you if asked kindly
Four of those five are solved by a polite message with the right detail attached. None of them are helped by waiting.
A sequence that works
Consistency does the work here, not escalation. The point of a sequence is that nobody has to decide, each time, whether to chase — the decision was made once.
| When | What goes out | Tone |
|---|---|---|
| 3 days before due | Friendly heads-up with the invoice attached again | Helpful. Solves the "lost it" case before it happens. |
| Day 1 overdue | Short note, invoice number, amount, payment details | Neutral, assumes an oversight. Usually is one. |
| Day 7 | Ask a question rather than repeat a demand: is it approved, is anything needed from us? | Curious. Flushes out the PO or query problem. |
| Day 14 | Named person, phone call, email follow-up confirming what was agreed | Personal. This is where most of the older debt actually clears. |
| Day 30 | Formal reminder, terms restated, statutory interest mentioned as fact | Firm, unemotional, still leaves the relationship intact. |
Two details matter more than the schedule. Always include everything needed to pay — invoice number, amount, bank details, a link if you have one — because every additional step is another day. And always send from a real person's name; "accounts@" gets triaged, a person gets answered.
UK businesses can charge statutory interest on late commercial payments, plus a fixed sum per invoice, under the Late Payment of Commercial Debts legislation. Most small firms never invoke it, and often that's the right call commercially. But stating in a day-30 letter that interest is chargeable — as a fact, not a threat — is a legitimate and surprisingly effective nudge. Check the current rate and your own terms before quoting figures, and take advice if a debt is heading anywhere formal.
What to automate, and what not to
The mechanical parts are ideal for software: knowing what's overdue and by how long, knowing who to write to, knowing what was sent last time, and drafting the next message with the right details in it. That's the bit that never happens reliably by hand.
What shouldn't be automated is the send button on anything sensitive. There's a version of this that emails every overdue customer on a fixed schedule with no human in the loop, and it will eventually send a stiff reminder to the client you had lunch with on Tuesday, whose payment is late because you agreed it could be. One of those does more damage than a month of chasing does good.
So the shape we build is: software finds them, drafts them, and puts them in a queue. A person spends four minutes on a Tuesday reading and sending. Accounts that need special handling get flagged and skipped automatically. The work stops being a task and becomes a two-minute review.
Make it visible, not dramatic
Overdue debt should appear in your weekly report as a matter of course — total, oldest, biggest, and the direction of travel. Not because a number on a page collects money, but because a slowly worsening trend is invisible month to month and obvious across eight weeks. By the time it's a cashflow emergency, the cheap interventions have expired.
It's also worth watching payment behaviour per customer rather than only in aggregate. A good customer whose payment times have crept from 30 days to 55 is telling you something about their own cash position, several months before it becomes your problem. That's useful intelligence for credit limits, for order sizes, and occasionally for deciding whether to take on more work for them.
Prevention, which is duller and better
- Get terms agreed in writing before starting. A payment conversation is much easier when it references something signed.
- Invoice the day the work is done. Batching invoices weekly adds days to every single one, for no reason.
- Ask who approves and how, at the start. Two minutes at the beginning saves three chases at the end.
- Make paying frictionless — direct debit for recurring work, a payment link for one-offs. Every barrier is a delay.
- Take deposits for anything large or bespoke. Nobody has ever regretted this.
Working out what late payment costs you
Most businesses know they're owed money and don't know what the lateness itself costs. It's calculable.
Take your average debtor days and compare it with your terms. If terms are 30 days and the reality is 47, you're financing your customers for 17 days on every invoice. Multiply your average daily sales by those 17 days and that's the working capital permanently tied up.
For a business invoicing £40,000 a month, that's roughly £22,000 sitting in other people's accounts at all times. Whether that costs you interest, an overdraft facility, or simply the ability to take on the next job depends on your situation — but it's never free.
The useful part: a systematic chase process that pulls 47 days back to 38 releases about £12,000 of cash permanently, for the cost of some software and four minutes a week. That's a better return than most things a small business can do with £12,000.
- Debtor days — and the direction it's moving. A slow worsening is invisible month to month and obvious across eight weeks.
- Percentage of invoices paid within terms — because an average hides the difference between everyone being slightly late and a few being very late. Those need completely different responses.
Watching payment behaviour per customer
Aggregate figures tell you there's a problem. Per-customer patterns tell you which one, and give you warning.
A customer whose payment times have crept from 30 days to 55 over six months is telling you something about their own cash position long before it becomes your bad debt. That's actionable intelligence: it's a reason to review the credit limit, to ask for a deposit on the next large order, or simply to be more prompt about chasing them specifically.
The pattern to watch for is deterioration relative to that customer's own history, not relative to your terms. A customer who has always paid at 45 days and still does is predictable and fine. One who has moved from 20 to 35 is the one to look at, even though they're still inside terms.
The awkward conversations, scripted
The reason chasing gets avoided isn't the admin — it's the discomfort. Having wording ready removes most of that.
When they've gone quiet entirely. Ask a question rather than repeating a demand: "I wanted to check whether invoice 1042 reached the right person — is there anything you need from us to get it approved?" It gives them a face-saving route to admit it's stuck internally, which is usually the truth.
When they're a customer you value. Say so explicitly: "You're one of our better customers so I didn't want this to become an issue — invoice 1042 is now 40 days out. Is there a problem at your end we can work around?" That sentence protects the relationship while being unambiguous about the ask.
When they've broken a promise to pay. Reference the commitment, not the emotion: "You mentioned this would go in the run on the 15th and it hasn't come through — has something changed?" Factual, hard to argue with, and it establishes that promises are being tracked.
When it's heading somewhere formal. Stop improvising and take advice. A badly worded threat can undermine a legitimate claim, and the wording at that stage matters more than the tone.
When to stop chasing and change the deal
Some customers are structurally slow and will not change. At that point the answer isn't more chasing — it's changing the terms of trade.
The options, roughly in order of how much relationship they cost: ask for a deposit on new work; move them to payment-on-delivery; reduce the credit limit so exposure is capped; require direct debit for recurring work; or decline the next order.
The calculation is simple and rarely done: a customer at 90 days on 8% margin may be costing you more in financed working capital than they contribute. Nobody enjoys that conversation, but doing the arithmetic at least makes it a decision rather than a drift.
The point
None of this is clever. That's rather the argument: the money isn't recovered by a brilliant negotiation, it's recovered by asking early, consistently, with the right details, every time — which is precisely the sort of unglamorous repetition that software should be doing while you get on with the work you're actually good at.