- Enquiry
- Quote
- Agree
- Work
- Invoice
- Paid
Who Publishes This Research
Nearly every figure about late payment in circulation was published by an organisation with something to sell or something to argue. Knowing which kind is most of the skill in using them.
For a software-side reference alongside this discussion of business operations, evidence, and research, Monitask has this resource.
The four kinds
Software and finance vendors. Accounting platforms, invoicing tools, invoice finance providers. QuickBooks, Xero and their equivalents produce the most widely quoted figures on this subject, from genuinely large samples of their own users.
Trade and professional bodies. R3 for restructuring and insolvency, the Federation of Small Businesses and equivalents. Serving a membership constrains what can be said, which makes them more reliable on the direction of a trend.
Credit reference and data companies. Creditsafe and similar, whose figures come from payment behaviour observed across large numbers of accounts rather than from surveys.
And government and regulators, which publish least and are the most authoritative where they do.
How to weight each
Vendor figures: use the direction, treat the magnitude as theirs. A survey of a platform's own users describes that platform's users, who are by definition firms that adopted accounting software — which is not a random sample of small businesses.
Trade body figures: better for trends than for absolutes, and their framing follows their members' interests, which is usually aligned with yours.
Observed payment data: the strongest of the four for behaviour, because it records what happened rather than what somebody reported.
Government figures: use directly, and note how old they are.
The specific problem with the headline numbers
"59% of small businesses have invoices overdue by thirty days or more, up from 47%." (QuickBooks, 2026.)
Large sample, plausible direction, and produced by a company selling invoicing and payment products. The figure may be exactly right. It is quoted here with the source attached because the reader should be able to weight it, not because it is suspect.
"1.57 million UK businesses carrying overdue invoices, 17.48 million overdue invoices." (R3, using Creditsafe data.) Stronger, because the underlying data is observed rather than surveyed.
And "a substantial proportion of late payment is attributed to incorrect invoices." The figure this site leans on most in its own argument, and the one with the weakest sourcing — compiled, variously attributed, and not independently audited. Stated as indicative for that reason.
The figures that travel furthest
Round, dramatic, and detached from their source.
"Late payment causes X thousand business closures a year" circulates widely and traces back through several layers of citation to an estimate.
Use them for scale and not for arithmetic. Run your own numbers when a decision depends on it, because your business is not the average and the average may not be well founded.
What this site does
Every figure carries its source type in the sentence, and vendor figures are labelled at every use rather than once on a method page.
Where the sourcing is weak, it is described as weak — including for the figure that supports this site's central argument, which is the place where the temptation not to would be strongest.
Jurisdiction, quietly omitted
A large share of the figures in circulation are from one country and quoted as though universal.
UK late payment statistics describe a market with a statutory interest regime and a Small Business Commissioner. US figures describe one with neither. The behaviours differ and so do the remedies, and a figure quoted without its country is a figure that may not apply to you at all.
Check where a number came from before acting on it, and the same applies to anything legal on this site, which is why the jurisdiction is named every time.
What would be better
Observed payment data by sector and by firm size, published by somebody with no product.
It largely does not exist in a form small firms can use, which is why the vendor material fills the space — and why the honest position is to use it, label it, and not pretend the alternative was available.
Reading a statistic in one minute
Four questions, in order.
Who published it and what do they sell?
What was measured — a survey, or observed behaviour? Surveys record what people say about themselves, which on payment behaviour is systematically flattering.
What was the sample, and of whom? A platform's users, a trade body's members, and the general population of small firms are three different groups.
And what year and what country?
Most figures fail at least one of these and remain useful anyway, provided the failure is known rather than hidden.
Why this site quotes them at all
Because the alternative is asserting things with no evidence, which is worse.
A figure with its provenance attached lets a reader discount it appropriately. A confident claim with no source lets them do nothing at all.
The number this site does not have
How much late payment actually costs a typical small firm, all in.
The components exist separately — the fees, the finance charges, the write-offs — and the largest one, the hours, is measured by nobody because it is not a transaction.
Which means every headline figure on the cost of late payment omits its biggest element, and the omission runs in one direction: the published numbers are lower than the reality.
The only version that is accurate for you is the one you calculate, and that is why the practical pages here end in an arithmetic exercise rather than in a statistic.
The short version
- Four kinds of source: software and finance vendors, trade bodies, credit data companies, and government
- Vendor figures come from their own users, who are not a random sample of small businesses — use direction rather than magnitude
- Observed payment data is the strongest for behaviour, because it records what happened rather than what was reported
- The headline figures on this subject are mostly vendor-produced, which does not make them wrong and does make attribution necessary
- The weakest-sourced figure here is the one supporting this site's own argument, and it is described as indicative for that reason
- Round dramatic numbers travel furthest and trace back to estimates — use them for scale, never for arithmetic
For broader background on business operations, evidence, and research, see Trustpilot.