What it is
In the two-corner model, an e-invoice travels from a single sender to a single receiver over a connection the two parties have agreed and configured ahead of time. There is no shared directory, no neutral intermediary, and no notion of "send to anyone, receive from anyone." Each pairing is its own integration project: agree the format, agree the transport, agree the validation rules, exchange test files, go live.
The two corners are simply the two endpoints: corner 1 is the supplier's billing system, corner 2 is the buyer's accounts payable system. Everything in between — VAN providers, secure FTP, AS2 / AS4 channels, even sneakernet of CSV files on a Friday afternoon — is treated as transport plumbing rather than as a structural part of the model.
Where it came from
The two-corner model is the original shape of B2B EDI. From the 1980s onwards, large buyers (especially retailers and automotive manufacturers) imposed an EDI mandate on their suppliers: "if you want to keep selling to us, here is the specification you must implement to send invoices, despatch advices, and remittance advices." The supplier built one integration per customer, often with a different format (EDIFACT INVOIC D96A here, ANSI X12 810 there, a buyer-specific XML somewhere else) and a different VAN account.
This worked when each large buyer could justify the integration cost on its own volume. It collapses as soon as a long tail of mid-sized buyers and small suppliers tries to participate: every new pairing is a project, costs aggregate quadratically with the number of trading partners, and onboarding a single supplier into a new buyer's network can take weeks.
Two corners vs. four corners
The four-corner model — the foundation of Peppol — solves the quadratic-cost problem by inserting a service provider on each side. Corner 1 is still the sender, corner 4 is still the receiver, but corner 2 (the sender's access point) and corner 3 (the receiver's access point) handle format translation, transport, and discovery. Each party connects once to their access point and can reach any other participant in the network.
From an ERP vendor's perspective the practical differences are:
A five-corner variant exists — used in CTC regimes like France's PPF model and Peppol's pilot 5-corner architecture for tax reporting — where a tax authority is inserted as the fifth corner alongside the four exchange parties.
Where the two-corner model still exists
It has not disappeared. It survives wherever the underlying economics still favour it:
The two-corner model also resurfaces as a fallback inside otherwise four-corner architectures: when a Peppol access point cannot resolve a recipient, the issuer is sometimes permitted to deliver by direct e-mail or upload portal, which is a two-corner exchange dressed in network terminology.
Why ERP vendors still care
Even for ERPs targeting fully four-corner customers, two-corner integrations remain in the backlog because most mid-sized buyers have at least one trading partner who refuses to join Peppol. A modern invoicing module typically needs to:
Relation to EN 16931
EN 16931 is corner-agnostic. The semantic data model says nothing about how the invoice gets from sender to receiver. A perfectly compliant EN 16931 UBL invoice can be delivered two-corner over SFTP, four-corner over Peppol AS4, or five-corner via France's PPF — the semantic content is the same.
The practical effect is that two-corner is a deployment model, not a format choice. ERPs aiming at EU readiness build their internal data path around EN 16931 and treat the corner count as a routing decision.