E-invoicing mandate 2027: rebuild your invoice flow

In a business an invoice is not a document you file but a case that passes through several hands. An incoming invoice is received, checked for accuracy, matched to an order or cost centre, approved, paid and archived. An outgoing invoice is created from an order, a delivery or a service rendered, gets the right line items, prices and tax rates, is sent out and has to be found again later. At each of these stations many companies still rely on manual work today — and that manual work, not the paper or the postage, is the real cost block.

The difference between a PDF and a structured e-invoice lies not in appearance but in machine readability. A PDF is a picture of a document; a program sees no amounts there, only pixels. A structured e-invoice to EN 16931 (EN 16931) contains the same details as machine-readable data fields: invoice number, date, supplier, line items, net, tax, payment terms. That lets the system read an incoming invoice itself and propose it to accounting instead of someone copying the figures. The entire benefit of the change hangs on this one point: data instead of a document image.

So for a business that treats it as a workflow topic, the 2027 deadline is good news rather than a burden. It forces you to look at the invoice path in full — from which system creates the invoice to how it reaches accounting without a second capture. Lay that chain out cleanly once and you gain an effect that goes beyond the obligation: fewer typing errors, faster approvals, cleaner figures. The jump from 64 million (DATEV) e-invoices across all of 2025 to over 51 million in the first half of 2026 alone is not an administrative detail but a sign of a switch that has arrived in mid-size companies.

Receiving is the part often overlooked because no deadline looms here — the duty to receive has applied for a while. That is exactly why the most unused potential sits here. In many companies an incoming XRechnung lands in the mail inbox, is opened, the figures are eyeballed and typed by hand into accounting or the ERP. That turns a machine-readable invoice back into manual work — the most expensive possible way to handle the new format. A clean inbound path takes the file automatically, reads the fields and presents them for checking. How far such routine back-office work can be taken off people's hands is covered in AI in the back office.

The first step is a fixed inbound channel instead of many personal inboxes. A central invoice address that collects incoming e-invoices is the basis for them being processed by machine at all. The second step is reading them out: from the structured file, supplier, invoice number, line items and amounts are taken over as a data record. The third step is the accuracy check — matching against order and delivery note — which only becomes automatable thanks to the available data. How a business builds such handovers without capturing twice is shown in eliminating duplicate data entry.

At the end of this path comes archiving, and for e-invoices that is no side point. The structured file is the original that counts for tax and must be kept unchangeably — a printout does not suffice, and leaving the original only in the mail inbox is not orderly filing. What to watch for is covered in GoBD-compliant document storage; the concrete periods are the topic of meeting retention periods digitally. For storing the machine-readable documents in the business, orderly document digitisation is the right frame.

The real gain appears at the seam between systems. An e-invoice that is created but then typed into accounting by hand again has missed half its purpose. What makes sense is a handover in which the invoice data flows from the leading system into accounting and produces a posting proposal there. Whether that happens over a direct interface, a structured handover file or a document transfer to the tax office depends on the setup — the goal is the same in every case: no second capture of the same figures.

The cost calculation behind it is simple. Every document captured by hand costs a few minutes and carries a risk of error; over a year and many documents that adds up to a fixed amount that recurs every month. A handover built once, by contrast, costs once and then carries without further manual work. How to run that calculation for your own business is shown in what a single case really costs and interface or manual work. For structured merging of invoice, order and payment data, clean data integration is the frame.

The larger part of a successful changeover lies in preparation, and that costs no money at first, only clarity. Four questions go a long way: which system creates the outgoing invoices today and what structured formats can it produce? Where do incoming invoices go and who processes them? In which format should outgoing invoices leave in future, measured against the recipients? And how does the data reach accounting or the tax office without a second capture? Settle those four points in 2026 and 2027 brings a routine, not a deadline.

It also helps to take an honest inventory of the media breaks. Often it turns out that the invoice path is only one of several where the same data is touched more than once — quote, order, delivery note, invoice. Where a change is due anyway, it is worth looking at the whole chain; spotting media breaks in your company helps with that inventory. And whatever runs on permanently after the change — looking after the invoice flow, small adjustments, handling faults — belongs in orderly operations and not in the remaining time of a project phase.