The Clearance Model
Mandatory e-invoicing in Europe, read at document level
Most e-invoicing projects fail on data rather than on technology. The mandate arrives, the software is bought, and then it turns out that the customer master has no reliable tax identifier for a third of the accounts, that the item master carries no unit of measure code, and that nobody can say which of the seven document types the accounts receivable system emits are actually invoices. This section is about the work between the decision and the first live document: readiness assessment, master data remediation, redesigning accounts payable around a document that arrives already structured, supplier and customer onboarding at scale, testing against a real validation stack, exception handling once volume is flowing, and the cost model that decides whether the work is done in the ERP, in a service provider's platform, or in both.
Scope, documents, data, systems and the receiving side, worked in the order the answers bind. Done properly it produces a cost range you can defend line by line.
The analysis underneath the anchor piece.
During a project, exceptions are worked by people with capacity and expertise. Three weeks after go-live they are not, and that is a capacity decision in technical clothing.
This gets framed as build or buy, and it is really a question about where knowledge sits. Most large businesses end up splitting it, and the split has a cost that nobody puts in the comparison.
These projects fail on data, not on technology. The fields a structured invoice makes load-bearing were decorative before, and the difference between present and correct is where the whole cost sits.
When the obligation to receive arrives, every supplier has to be able to reach you. Contacting them all and asking for their details does not survive contact with a supplier base of any size.
On the receiving side a mandate deletes steps rather than adding them, and the control that used to sit inside those steps has to be put back somewhere deliberately.
Four gates stand between a document and acceptance and only three are published. Passing the public ones and failing the fourth in production is the normal project outcome.
Licence and transaction fees are the visible cost and rarely the largest one. A business case built on price per document will be wrong in a direction that is discovered late.