Your AP metrics were built for the wrong suppliers
The following is a guest post from Boris Dzhingarov, CEO at ESBO Ltd.. Opinions are the author’s own.
Accounts payable benchmarks assume a company pays a manageable number of suppliers. Ardent Partners, in its State of ePayables 2025 research, puts the average cost of processing a single invoice at $9.84 and the average processing time at 8.2 days. Those figures are useful if your payables run through a few hundred vendors who quote a purchase order number and accept a bank transfer. They tell you very little if your vendor list runs into five figures.
Mine does. I run ESBO Ltd., a digital PR and link building agency based in Plovdiv, Bulgaria. There are about 20 of us. We pay more than 10,000 publishers, nearly all of them one- or two-person operations, in different countries and different currencies. By revenue and headcount, we are a small business. By supplier count, our payables function resembles that of a company many times our size, and no amount of software has changed that.
The shape of the problem is spreading. Marketing, content, logistics and professional services have all pushed spend toward large numbers of very small vendors: freelancers, creators, single-site publishers, regional subcontractors. Finance teams usually inherit that fragmentation after the commercial decision has been made, without extra AP headcount, and then benchmark themselves against numbers that were never built for it.
Regulation is about to make the gap expensive. The European Union’s VAT in the Digital Age package entered into force in April 2025. Digital reporting for cross-border business-to-business transactions becomes mandatory on July 1, 2030. Domestic systems must align with the EU standard by January 2035, and in the meantime member states are free to impose their own e-invoicing mandates. Several already have, and more come online each year. Ardent Partners found that 57% of suppliers can currently send an invoice electronically. That is a workable number when the other 43% is a few dozen firms someone on your team can phone. It is a different situation when it is 4,000 sole traders who have never encountered a structured invoice format.
We learned four things the expensive way.
1. The unit of cost is the supplier, not the invoice. Cost per invoice measures keystrokes. What actually consumes the team is the relationship around it: identity checks, tax status, bank details and the email thread that starts when a payment bounces because a beneficiary name does not match an account. Ardent Partners found that AP staff spend 21.9% of their time dealing with suppliers, and at our volume that share is much higher. A vendor who invoices once a year costs nearly as much to carry as one who invoices monthly. Our forecasting only became useful when we budgeted AP capacity against supplier count and new suppliers added per month rather than invoice volume.
2. Supplier enablement targets fail on the tail. The standard advice is to move suppliers onto your portal or your format. Large vendors comply because your spend matters to them. A publisher earning a few hundred euros a year from you will not, and chasing them costs more than the invoice is worth. We stopped asking. We accept whatever they send and do the conversion on our side. That is the opposite of what most automation vendors recommend, and for the long tail it is the only approach that survives contact with reality.
3. Payment fragmentation costs more than payment fees. Every additional currency and rail creates its own reconciliation path. The visible cost is the transfer fee. The real one is the hours spent matching a payment that landed short because an intermediary bank took a cut, or clearing FX differences across hundreds of small transfers. We cut this by reducing the number of rails we support, not by negotiating better rates.
4. Fraud moves to the vendor master. When you have thousands of counterparties you have never met, the exposure is not a fabricated invoice. It is a change of bank details on a genuine supplier account, requested by email, by someone whose voice nobody at the company has ever heard. We verify banking changes on a fixed schedule rather than treating each request on its merits, because at this volume the request itself carries no signal.
For CFOs at larger organizations, the practical question is how much of your spend already sits in that tail, and whether anyone in the building can tell you. In most vendor masters, the answer is hidden, because reporting is organized by spend. Sort by supplier count instead and the picture inverts: a rounding error in euros, the majority of your records and most of your compliance surface. The 2030 deadline applies to the records, not to the money.
The work that follows is slow and unglamorous. Deduplicate the vendor master. Verify tax identifiers. Decide which small suppliers to consolidate behind an intermediary and which to keep direct. Accept that a share of your vendors will never send a structured invoice and design for that instead of legislating against it. Anyone planning to start this in 2029 will find that the binding constraint is not the system. It is how long it takes to get several thousand very small counterparties to answer an email.
