The question
What does an order pass through between the sale and the money, and where do the time and the money go?
Method 37 published sources from 2001 to 2026, each read on its primary release and marked with who ran it and what it counted. Official statistics and peer-reviewed studies first; two vendor surveys, named as such.
A sale closes in one place and is finished in several others. Before the money arrives, the same order becomes an invoice, a payment to match and, where goods are involved, a purchase from a supplier, a receipt and a delivery. Around 18 billion invoices are issued in the European Union every year, more than 500 a second1, and each stands for an order that passed from one person, or one company, to the next.
We read 37 published sources for what is known about those handovers: 18 peer-reviewed studies, 13 official sources, one working paper, three studies by companies of their own data and two vendor surveys. Beside every reference we say what kind of evidence it is and what it counted. No study follows one order from the sale to the money, so the piece goes handover by handover, and says where the evidence comes from shops, from large listed firms or from another field altogether. Every number below is taken from its primary release; the selection, and what we make of it, are ours.
Key findings
Most firms run an order without one shared record. In 2025, 46.45% of EU enterprises with ten or more people used ERP software, the kind that shares information between functions. Among those with 10 to 49 people the share was 41.08%2.
Nobody has counted how often an order is typed again, but the cost of typing is measured. An invoice that arrives as a PDF by email costs its receiver EUR 7.68 to process, against EUR 8.04 on paper and EUR 2.27 as structured data3. In 2023, 38.68% of EU enterprises sent any invoice a machine can read4.
The average invoice is paid near its date. The slow ones are the problem. Large Australian businesses pay small suppliers in 27.2 days on average against a 29-day term, yet it takes 55 days for 95% of those invoices to be paid5. In the United Kingdom, large businesses paid 16% of invoices late in 20246.
Each side says the other pays late. In a European Commission survey, 73% of respondents said their customers pay them late, and 24% admitted to paying late themselves7.
Chasing takes real hours, and two counts of them are six times apart. A vendor survey puts it at 9.85 hours a week for each company8. A government-commissioned survey in the United Kingdom puts it at 86 hours a year for each business that chases at all9.
Where stock records have been counted, most were wrong. 65% of nearly 370,000 records in 37 stores of one retailer did not match the shelf10. A 2025 working paper on another retailer found 35.3% accurate11.
A missed promise is paid for in later orders. At one supplier, a one-point rise in the share of ordered goods it could ship went with an 11% rise in what its business customers ordered next12.
How many handovers does one order pass?
Several, and no published source counts them all. An invoice "is usually preceded by the exchange of a purchase order and various fulfilment/transport documents", in the words of a report for the European Commission13. A research brief from the US Federal Reserve divides the work between two trading companies into five phases: taking on a supplier, ordering and delivery, invoicing, payment, and matching the payment to the invoice14.
The same brief names three points where one order's details may be typed again by hand. The supplier types in a purchase order that arrives in a form its system cannot import. The buyer scans or types an invoice that arrives on paper or as a PDF. The supplier then matches the payment to a remittance note that "may arrive hours or days earlier or later via a completely different channel"14. The brief gives no count of how often each happens. We found no peer-reviewed or official source that does.
What official statistics do count is how many firms have one system that shares an order between functions. In the European Union in 2025, 46.45% of enterprises with ten or more people used ERP software, which Eurostat defines as software "to share information between different functional areas". The share was 41.08% among enterprises of 10 to 49 people, 69.93% among those of 50 to 249, and 88.71% among those of 250 or more2. Fewer still exchange that information with the companies they trade with: 24.61% shared supply chain information electronically with suppliers or customers in 20234.
Two cautions belong beside these numbers. Having such software is not the same as sales, accounts and the stockroom reading one record; it is the nearest thing official statistics measure. And enterprises with fewer than ten people are not surveyed at all.
Why a handover costs time is easier to show than to count. In one manufacturing process studied from its event log, work was being done during 6.81% of the time a case spent in the process. The largest single cause of waiting, 57% of it, was that the next person was not available when the work reached them15. That is one process in one factory, an illustration of the mechanism and not a rate for orders in general.
What happens when an order is typed again?
Mistakes, at a rate measured in another field. We found no study of error rates in typing orders or invoices. The closest measurement is from clinical research, where trained staff enter data from paper forms. A review that pooled 93 papers found errors in 0.29% of fields when the data was typed once and in 0.14% when it was typed twice and compared. When staff copied the data out of medical records, which is reading a document and not typing a form, the pooled error rate was 6.57%16. Reading an order out of an email and typing it into another system resembles the last of these more than the first, but nobody has measured it.
Most invoices still arrive in a form that someone has to read. In 2023, 38.68% of EU enterprises sent invoices suitable for automated processing, 72.84% sent electronic invoices that are not, such as a PDF by email, and 70.24% sent invoices on paper. A firm can do all three, so the shares overlap4.
The difference shows in what an invoice costs the company that receives it. A study of Belgian businesses put the cost of receiving one business invoice at EUR 8.04 on paper, EUR 7.68 as a PDF by email and EUR 2.27 as structured data3. The PDF saves the sender a stamp. It saves the receiver almost nothing, because someone still types it in.
Figure 2
A PDF invoice costs its receiver almost what a paper one does
Cost of receiving one business invoice, by how it arrives
| Measure | Value |
|---|---|
| On paper | 8.04 EUR |
| As a PDF by email | 7.68 EUR |
| As structured data | 2.27 EUR |
Payment has its own paper. In the United States, checks fell from 32% of payments between businesses in 2015 to 13% in 2024 by number, while their total value barely moved, from USD 15.4 trillion to USD 15.0 trillion17. The central bank's researchers note that with a check both sides "must handle several business processes manually".
- What one shared system changed: timeAt one company followed before and after it moved to a single system, the time from order to delivery improved at once and went on improving, where it had been flat before18.
- What it changed: profitAcross 186 announced moves to such systems at listed firms, there was some evidence of better profitability and none of better share returns. For 80 moves to CRM software alone, there was no evidence of either19.
How long does an invoice wait?
Asked, companies say weeks past the date agreed. Measured, the average is close to the date and the slow invoices are very slow. In a credit management company's survey of 9,150 executives in 25 European countries, suppliers report an average term of 43 days for business customers and payment after 60 days. For public authorities the term is 55.9 days and payment comes after 69.878. These are executives' estimates, not counted invoices.
Figure 3
Suppliers say they are paid more than two weeks past the date agreed
Days agreed and days it took to be paid, by who pays, as suppliers report it
| Measure | Agreed | Paid |
|---|---|---|
| Business customers | 43 days | 60.3 days |
| Public authorities | 55.9 days | 69.8 days |
Two governments make large businesses report how they actually pay. In the United Kingdom, the typical large business took 32 days to pay in 2024, down from 35 in 2018, and paid 16% of its invoices late, down from 26%. In every year from 2018 to 2024, between 7% and 10% of invoices took longer than 60 days6.
In Australia, 3,254 large businesses reported on the second half of 2025. They paid small suppliers in 27.2 days on average against an agreed 29, and 68.5% of invoices were paid within terms. But 24.3% of invoices took 31 to 60 days and 5.9% took longer, and it took 55 days for 95% of invoices to be paid. The regulator's own summary: "It is taking almost double the average agreed payment term for the vast majority of small business invoices to be paid"5.
Figure 4
Of a hundred invoices to small suppliers, thirty wait more than thirty days
Invoices from small suppliers, by how long large businesses took to pay
| Measure | Value |
|---|---|
| Paid within 30 days | 69.8% |
| Paid in 31 to 60 days | 24.3% |
| Paid after more than 60 days | 5.9% |
The same wait can be read from the seller's side. In the invoices that small businesses issue through one accounting software company, customers took 29.5 days to pay in the United Kingdom in the three months to September 2026 and paid 8.0 days late on average. In the United States the figures were 29.7 and 8.6 days, in Australia 24.3 and 6.920. A longer series from another accounting software company, 58 million invoices from 110,000 small UK businesses, shows the wait falling from 81 days in 2010 to about 36 in 2020 and 2021, with the share paid after the due date going from about 47% to 40%21.
In a US central bank survey of small employer firms, 39% named slow-paying customers as a difficulty with payments, second only to processing fees22.
Why is an invoice paid late?
Each side points at the other, and some of the delay is paperwork. In a European Commission survey with 3,500 responses, 73% said their customers pay them late, and 24% admitted to paying late themselves7. Both cannot describe the same invoices.
Figure 5
Three in four say they are paid late. One in four says it pays late
Share of respondents who say...
| Measure | Value |
|---|---|
| their customers pay them late | 73% |
| they pay late themselves | 24% |
- The customer is short of cashIn a trade credit insurer's survey of 210 UK firms, 51% named their customers' cash flow as a reason for late payment23.
- The invoice is stuck inside the customerIn the same survey, 23% named internal approval delays, 26% banking delays and 17% a complex payment process23. In Austria, 43% of companies named administrative inefficiency as a driver of late payment between businesses in 20247.
- Paying late is cheaper than borrowingIn the UK government's research, 18% of businesses put being paid late down to customers using it as interest-free finance6.
- The seller agreed to it55% of European companies say they accept longer payment terms than they are comfortable with, so as not to damage the relationship with a client24.
Lateness also travels. In 2024, 31% of European companies said they delay their own payments because they are paid late themselves7.
A cleaner invoice removes the paperwork part and not the rest. Missing details or a wrong invoice number can stop a payment from being matched to its invoice without a person14. But a report for the European Commission concludes that structured invoices have "great potential to improve payment efficiency" and do not by themselves change payment behaviour: a buyer who never meant to pay on time will not start13. No source we found measures how many days an invoice error or a dispute adds.
What does the wait cost?
Hours first. How many depends on who counted. The credit management company's survey puts the time spent chasing late payments at 9.85 hours a week for each company, down from 10.15 the year before8. The official EU report on payments repeats that figure7. A survey of 1,455 businesses for the UK government found that 22% spend staff time chasing late payments, on average 86 hours a year for each business affected, 133 million hours in all9. Eighty-six hours a year is under two hours a week. The first figure comes from a company that sells collection services, and the second counts only businesses that said they chase.
Figure 6
Two counts of the time spent chasing payment, six times apart
Hours a week spent chasing late payments, for each company
| Measure | Value |
|---|---|
| A credit management company's survey, 25 European countries | 9.85 h |
| A government-commissioned survey, United Kingdom | 1.65 h |
The same UK research puts the whole cost of late payment at almost GBP 11 billion a year: GBP 4.60 billion in recovering unpaid invoices, GBP 3.21 billion in business closures, GBP 1.71 billion in investment not made and GBP 1.21 billion in raising other finance6. Its estimate of 14,000 closures a year, 38 a day, rests on a matched sample of 300 firms9. These are modelled estimates, and the closure figure is the least certain of them.
Peer-reviewed studies of what the wait does to a firm are fewer and firmer.
- Shorter terms, fewer failuresWhen a French reform capped the payment terms trucking firms could be made to give, their probability of default fell by 25%, and more small firms entered the trade25.
- Faster payment, more hiringWhen the US government began paying its small contractors 15 days sooner, their payroll rose by 10 cents for each dollar paid early. The published study adds that across a whole local labour market the gain was close to zero where labour was already scarce26.
- A customer's failure spreadsIn Sweden, a supplier whose customer went bankrupt owing it money saw its own annual risk of failure rise by around 2 percentage points, from an average of 2%27.
- Late payment alone, a limited effectThe French central bank found that late customer payments raise a company's probability of default by 25%, and by 40% when more than a month late. It calls the effect "relatively limited": 8 in 100 failing companies were exposed to it, and weak finances raised the probability at least four times28.
How long a firm can wait depends on what it holds. In the accounts of 597,000 small US businesses at one bank, the median held enough cash to cover 27 days of its outgoings29. The invoices above took about 30 days to be paid.
How often is the stock record wrong?
Where it has been counted, more often than it is right. A promise to a customer is made from a record of what is in stock, and the record can be wrong. In the largest published count, researchers compared nearly 370,000 stock records in 37 stores of one retailer with what was on the shelves and found 65% of them inaccurate10. A working paper from 2025 on 11 stores of a UK grocer found almost the same: 35.3% of records matched the shelf, 39.4% showed more than was there and 25.3% showed less11.
Figure 7
Of a hundred stock records, thirty-five matched the shelf
Stock records in 11 stores of one grocer, against a count of the shelf
| Measure | Value |
|---|---|
| Record matches the shelf | 35.3% |
| Record shows less than is there | 25.3% |
| Record shows more than is there | 39.4% |
A record that shows more than the shelf holds is the costly kind, because nothing reorders an item the system believes is there. When the grocer's stores were counted and their records corrected, sales rose by about 11% across the store, all of it on items whose record had shown more than was there11. An earlier study of two retailers put the cost of wrong records and misplaced goods at more than 10% of profit, and found that performance differed widely between stores of one chain using the same software30. In field experiments at another retailer, tracking goods by radio tag cut inaccuracy by about 26%31.
All of that is from shops. For companies that sell to other companies we found no count of how many records are wrong. The nearest is a study of a parts distribution centre and an engineering wholesaler, where orders are accepted on the system's record alone. It describes a company that promises a sale on the strength of the record and goes on "to find out that it is unable to deliver when the shipment is due". The two companies logged 1,083 and 1,532 corrections to stock over 25 months32.
What does a missed promise cost?
Later orders. The clearest evidence between businesses is a field experiment at one fashion supplier. A rise of one percentage point in the share of ordered goods it could ship, measured over the year before, went with an 11% rise in what its retailer customers ordered next12.
Promising more than can be delivered works once. An online retailer showed some customers a faster delivery promise without changing how fast it delivered. Sales and profit rose, and so did returns, and fewer of those customers came back33. At another online retailer, previously loyal customers ordered less often and in smaller amounts after an order arrived late34.
When a delivery fails, the cause is more often inside the company than at its supplier. Of 827 supply disruptions announced by listed firms between 1989 and 2000, the firm's own failure was named as responsible in 278, a supplier in 120 and a customer in 10635. The reasons named most often were parts shortages, problems starting up a new product, customers changing their orders and production problems. In a related sample of 885 such failures, the firms reported 6.92% lower sales growth and 10.66% higher growth in costs than comparable firms36. These are large listed companies whose failures were serious enough to announce.
Figure 8
A failed delivery starts inside the company more often than at its supplier
Supply disruptions announced by listed firms, 1989 to 2000, by who was responsible
| Measure | Value |
|---|---|
| The firm itself | 278 |
| A supplier | 120 |
| A customer | 106 |
Why sales and operations fall out of step has been studied closely in one company. Sales, finance and operations each plan their own part, "which results in conflicts over expectations, preferences, and priorities". The study's conclusion is that agreeing on what will actually be done can matter more than the quality of the information each side holds37.
What this does not show
- No study follows one order. Each handover was studied by different people, in different countries, industries and decades. Set side by side they describe a path. They do not measure it.
- The retyping is not counted. We found no source that counts how many times an order is typed again, and none that measures typing errors in orders or invoices. The error rates here come from clinical research data.
- The stock evidence is from shops. Every share of wrong records is from retail stores. For companies that sell to other companies there are two case studies and no share.
- The delivery evidence is from a few companies. One fashion supplier, two online retailers, and large listed firms between 1989 and 2000. None is an ordinary late delivery at an ordinary company.
- The best payment data covers large payers. The measured figures are large businesses reporting on themselves in two countries, and the customers of two accounting software companies. The European averages are executives' estimates in a vendor's survey.
- The costs are estimates. GBP 11 billion a year and 38 closures a day come from a model, the closures from a matched sample of 300 firms.
- Three numbers are from working-paper versions. For the studies of faster payment, of customer failure and of supply disruptions, we read the figures in the authors' earlier versions; the published texts may differ slightly.
- The smallest firms are missing. The EU statistics cover enterprises with ten or more people.
- Two sources are vendor surveys, and we sell software. Both come from companies that sell credit services to the firms they asked. We sell business software. Read their numbers, and our reading of all of them, with that in mind.
What we build from this
Each need below is one the evidence names. Beside it is what SalexHub does about it today, with the page that shows it. Where we have nothing to show, the row says so.
An order that is not typed again
The Agent drafts the invoice from the quote your customer accepted, with the same lines and total, naming its quote. Invoicing in parts, the quote counts what is left. You issue it.
A promise made from what is really there
Confirming a customer's order holds the goods for it, so two orders cannot promise the same stock. A deal shows which of its lines are in stock and which have to be bought.
To know the day an invoice goes past due
A Mini, a small agent with one job, watches the sent invoices and reports the day one goes past due, with what is owed and what happened last on the account.
Fewer hours chasing
Ask who is late paying. The Agent finds the invoices past their due date and writes one reminder for each, from your own address. You edit and send.
A payment matched to its invoices
Tell the Agent what arrived. It splits the payment across the invoices it pays, and each reads Paid or Partially paid. Recording money always waits for your yes.
A stock record with its reasons
Every stock figure opens to the movements behind it, each with the document that made it. A count is entered as an adjustment with its reason, and an Owner or Admin posts it.
A late supplier, known the same morning
A Mini checks each weekday morning which purchase orders were due the day before and have not arrived, and proposes a call to the supplier.
Invoices a machine can read, and payment from the invoice
SalexHub does not send invoices through an e-invoicing network, and a customer cannot pay an invoice online.
Nothing to show yet
Proof in numbers
We have not measured what any of this saves in hours or in days to payment. Until we do, we make no claim about it.
Nothing to show yet
References
Grouped by the weight of the evidence, the heaviest first. Open a group to see its sources.
Peer-reviewed studyPublished in a journal after review by other researchers.18
- 03Assessing the electronic invoicing potential for private sector firms in Belgium
Poel, Marneffe and Vanlaer. The International Journal of Digital Accounting Research, 16, pages 1 to 34, 2016.
Who was askedThree annual surveys of Belgian private sector businesses on their invoicing of 2012 to 2014, the largest with 847 respondents.
Back to the text - 10Inventory Record Inaccuracy: An Empirical Analysis
DeHoratius and Raman. Management Science, 54(4), pages 627 to 641, April 2008.
Who was askedNearly 370,000 stock records in 37 stores of one US retailer, compared with a count of the shelves.
Back to the text - 12The Impact of Supplier Inventory Service Level on Retailer Demand
Craig, DeHoratius and Raman. Manufacturing & Service Operations Management, 18(4), pages 461 to 474, October 2016.
Who was askedA field experiment at one fashion supplier and the orders of its retailer customers.
Back to the text - 15Why Am I Waiting? Data-Driven Analysis of Waiting Times in Business Processes
Lashkevich, Milani, Chapela-Campa, Suvorau and Dumas. Advanced Information Systems Engineering (CAiSE 2023), Lecture Notes in Computer Science, pages 174 to 190, 2023.
Who was askedThe event log of one manufacturing process: 225 cases, 4,503 pieces of work, 46 people and machines. Numbers read in the authors' preprint.
Back to the text - 16Error rates of data processing methods in clinical research: A systematic review and meta-analysis of manuscripts identified through PubMed
Garza, Williams, Ounpraseuth, Hu, Lee, Snowden, Walden, Simon, Devlin, Young and Zozus. International Journal of Medical Informatics, 195, 105749, March 2025.
Who was asked93 papers published from 1978 to 2008 on how accurately data is entered in clinical trials.
Back to the text - 18Order Lead-Time Improvement following Enterprise Information Technology Implementation: An Empirical Study
Cotteleer and Bendoly. MIS Quarterly, 30(3), pages 643 to 660, September 2006.
Who was askedOne company, followed before and after it put in a single system. Read in the abstract, which gives no figures.
Back to the text - 19The impact of enterprise systems on corporate performance: A study of ERP, SCM, and CRM system implementations
Hendricks, Singhal and Stratman. Journal of Operations Management, 25(1), pages 65 to 82, January 2007.
Who was askedAnnouncements by listed firms: 186 of ERP systems, 140 of supply chain systems and 80 of CRM systems.
Back to the text - 25Trade Credit and Industry Dynamics: Evidence from Trucking Firms
Barrot. The Journal of Finance, 71(5), pages 1975 to 2016, October 2016.
Who was askedFrench trucking firms before and after a reform that capped the payment terms they give.
Back to the text - 26The Employment Effects of Faster Payment: Evidence from the Federal Quickpay Reform
Barrot and Nanda. The Journal of Finance, 75(6), pages 3139 to 3173, December 2020.
Who was askedSmall business contractors of the US government after a 2011 reform. The 10 cents figure is from the authors' working paper, NBER Working Paper 22420.
Back to the text - 27Trade Credit and the Propagation of Corporate Failure: An Empirical Analysis
Jacobson and von Schedvin. Econometrica, 83(4), pages 1315 to 1371, July 2015.
Who was askedSwedish companies and what they were owed when a customer went bankrupt. Figures read in the authors' working paper, Sveriges Riksbank Working Paper 263.
Back to the text - 30Execution: The Missing Link in Retail Operations
Raman, DeHoratius and Ton. California Management Review, 43(3), pages 136 to 152, Spring 2001.
Who was askedTwo retailers, one studied for wrong stock records and one for misplaced goods. Read in the abstract.
Back to the text - 31RFID-Enabled Visibility and Retail Inventory Record Inaccuracy: Experiments in the Field
Hardgrave, Aloysius and Goyal. Production and Operations Management, 22(4), pages 843 to 856, July 2013.
Who was askedTwo field experiments at one retailer: daily counts in 13 stores over 23 weeks, then five product categories in 62 stores.
Back to the text - 32Modeling (and Learning from) Inventory Inaccuracies in E-retailing/B2B Contexts
Rekik, Syntetos and Glock. Decision Sciences, 50(6), pages 1184 to 1223, December 2019.
Who was askedTwo companies that sell to other companies, a parts distribution centre and an engineering wholesaler, 25 months of data each. Read in the accepted manuscript.
Back to the text - 33Sooner or Later? Promising Delivery Speed in Online Retail
Cui, Lu, Sun and Golden. Manufacturing & Service Operations Management, 26(1), pages 233 to 251, January 2024.
Who was askedA field experiment at one US online retailer: the promised delivery time was changed and the real one was not. Read in the abstract, which gives no figures.
Back to the text - 34Failure to deliver? Linking online order fulfillment glitches with future purchase behavior
Rao, Griffis and Goldsby. Journal of Operations Management, 29(7-8), pages 692 to 703, November 2011.
Who was askedThe order records of one online retailer of printed material, for previously loyal customers. Read in the abstract, which gives no figures.
Back to the text - 35An Empirical Analysis of the Effect of Supply Chain Disruptions on Long-Run Stock Price Performance and Equity Risk of the Firm
Hendricks and Singhal. Production and Operations Management, 14(1), pages 35 to 52, March 2005.
Who was asked827 supply disruptions announced by firms listed on US stock exchanges, 1989 to 2000. The counts by cause were read in the authors' working-paper version.
Back to the text - 36Association Between Supply Chain Glitches and Operating Performance
Hendricks and Singhal. Management Science, 51(5), pages 695 to 711, May 2005.
Who was asked885 supply failures announced by listed firms, each compared with firms of similar size and industry.
Back to the text - 37Cross-functional alignment in supply chain planning: A case study of sales and operations planning
Oliva and Watson. Journal of Operations Management, 29(5), pages 434 to 448, July 2011.
Who was askedA case study of planning between sales and operations at one company. No statistics.
Back to the text
Official sourceStatistics or rules published by a public authority.13
- 01Proposal for a Regulation on combating late payment in commercial transactions
European Commission. COM(2023) 533 final, explanatory memorandum, 12 September 2023.
Who was askedThe European Union; the Commission's own count of invoices issued in a year.
Back to the text - 02Integration of internal processes by size class of enterprise
Eurostat. Dataset isoc_eb_iip, survey on ICT usage and e-commerce in enterprises, 2025 data, updated 27 February 2026.
Who was askedEU enterprises with 10 or more persons employed, all activities except agriculture, mining and the financial sector.
Back to the text - 04Integration with customers and suppliers, supply chain management by size class of enterprise
Eurostat. Dataset isoc_eb_ics, survey on ICT usage and e-commerce in enterprises, 2023 data, updated 17 July 2025.
Who was askedEU enterprises with 10 or more persons employed, all activities except agriculture, mining and the financial sector.
Back to the text - 05Regulator's update, August 2026
Payment Times Reporting Regulator, Australia. Commonwealth of Australia, 24 August 2026.
Who was asked3,254 large businesses reporting how they paid small business suppliers from 1 July to 31 December 2025.
Back to the text - 06Late Payments: Primary Legislation. Final stage impact assessment
UK Department for Business and Trade. Impact assessment DBT-054-25-DIME, March 2026.
Who was askedThe payment reports large UK businesses must file, 2018 to 2024, as medians across businesses; and the department's research on costs.
Back to the text - 07Annual Report 2025
EU Payment Observatory. Prepared for the European Commission by CEPS and EY; Publications Office of the European Union, December 2025.
Who was askedCompanies in the EU in 2024. It re-reads an official survey of about 11,000 firms, a vendor's survey of about 9,000 and a Commission survey with 3,500 responses.
Back to the text - 09Late Payments Research: Estimating the total economic cost of late payments and their impact on the UK economy
London Economics, for the UK Department for Business and Trade. Department for Business and Trade and the Office of the Small Business Commissioner, July 2025.
Who was asked1,455 UK businesses of all sizes, surveyed from 15 January to 7 February 2025; the closure estimate rests on a matched sample of 300 firms, 2015 to 2019.
Back to the text - 13How electronic invoicing helps reduce late payments in commercial transactions
EU Payment Observatory. 3rd Thematic Report, prepared for the European Commission by CEPS and EY; Publications Office of the European Union, July 2024.
Who was askedDesk research, data analysis and interviews. The report reflects the views of its authors.
Back to the text - 14B2B Payments: Business Processing and Challenges to Achieving Straight-Through Processing
Albers, Baird, Hayashi, Isaacson and Routh. Payments Research Brief, Federal Reserve Bank of Cleveland, 25 September 2026.
Who was askedA description of how US businesses order, invoice, pay and reconcile. It gives no counts.
Back to the text - 17B2B Payments: A Gradual Shift from Checks to Electronic Payment Methods
Albers, Baird, Hayashi, Isaacson and Routh. Payments Research Brief, Federal Reserve Bank of Cleveland, 21 September 2026.
Who was askedThe authors' estimates of payments between US businesses, 2015 to 2024, from the Federal Reserve Payments Study and other counts.
Back to the text - 222024 Report on Payments: Findings from the 2023 Small Business Credit Survey
Federal Reserve Banks. Small Business Credit Survey, December 2024.
Who was asked4,858 US employer firms with 1 to 499 employees that answered the question, surveyed from September to November 2023.
Back to the text - 24Annual Report 2024
EU Payment Observatory. Prepared for the European Commission; Publications Office of the European Union, December 2024.
Who was askedCompanies in the EU in 2023, re-read from official and vendor surveys.
Back to the text - 28Do late customer payments impact companies' probability of default?
Dietsch and Gonzalez. Bulletin de la Banque de France, 227/8, January and February 2020.
Who was askedFrench non-financial companies, 2014 to 2017, from balance sheet data.
Back to the text
Working paperResearch released before peer review.01
- 11Inventory record inaccuracy in grocery retailing: Impact of promotions and product perishability, and targeted effect of audits
Rekik, Oliva, Glock and Syntetos. arXiv:2506.05357, 2025, revised 2026.
Who was askedAbout 150,000 stock counts of some 24,000 products in 11 stores of a UK grocery retailer.
Back to the text
Company researchA company's study of its own data.03
- 20Xero Small Business Insights: United Kingdom, United States and Australia, September quarter 2026
Xero. Xero Small Business Insights, 29 October 2026.
Who was askedInvoices that small businesses issued through the company's software and that were paid in the quarter. The number of firms is not given.
Back to the text - 21Shedding light on late payments for UK small and micro businesses
Sage and Smart Data Foundry. Sage, with Smart Data Foundry at the University of Edinburgh, November 2022.
Who was asked58 million sales invoices from 110,000 small UK businesses using the company's software, 2010 to 2021.
Back to the text - 29Cash is King: Flows, Balances, and Buffer Days. Evidence from 600,000 Small Businesses
Farrell and Wheat, JPMorgan Chase Institute. JPMorgan Chase Institute, September 2016.
Who was asked597,000 small US businesses with accounts at the bank, February to October 2015.
Back to the text
Vendor surveyA survey run or paid for by a company that sells to the people it asked.02
- 08European Payment Report 2025
Intrum. Intrum, April 2025.
Who was asked9,150 executives in 25 European countries, surveyed from 14 November 2024 to 12 February 2025; 70% from companies with fewer than 250 employees.
Back to the text - 23Payment Practices Barometer: Survey results for United Kingdom, 2026
Atradius. Atradius, April 2026.
Who was asked210 UK companies of all sizes, interviewed online between the end of the first quarter and the start of the second quarter of 2026.
Back to the text
- Contains public sector information licensed under the Open Government Licence v3.0. It comes from the UK Department for Business and Trade's impact assessment of March 2026 and the Late Payments Research it published in July 2025, © Crown copyright 2026 and 2025. The licence: https://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/
- Regulator's update, August 2026, of the Payment Times Reporting Regulator: © Commonwealth of Australia 2026, licensed under a Creative Commons Attribution 4.0 International licence: https://creativecommons.org/licenses/by/4.0/. Based on Commonwealth of Australia data: the figure drawn from it is our own.
- The European Commission's proposal COM(2023) 533 and the EU Payment Observatory's reports of 2024 and 2025: © European Union, 2023 to 2025. Reused with the source acknowledged, under Commission Decision 2011/833/EU.
- Eurostat data (isoc_eb_iip, isoc_eb_ics): © European Union. Source: Eurostat. The values are printed as published, in a figure of our own.
- No source listed here took part in this piece, and none endorses SalexHub or its products.
Changes
- 30 October 2026First published.
