← All guides · EN 16931 & validation
My e-invoice was rejected: how to read the error and fix the real cause
By the Ingenious AI team · Published 14 November 2026 · Reviewed 14 November 2026
Convert an invoice free
Put this guide into practice now: convert one invoice free — validated UBL 2.1 XML plus a Factur-X PDF, with the acceptance checker for your client’s country. See pricing for bulk conversion, country rule packs and monitored clients.
Short answer
When an e-invoice is rejected, work the error list in dependency order: presence failures first (missing number, dates, parties, lines), then codes (type code, currency), then VAT category-rate agreement, then cross-border identifiers, then totals. Almost every rejection is a template or client-record defect wearing one invoice as a disguise — fix the source, re-validate here before resending, and the same rejection never happens twice.
What does a rejection mean?
A rejection means the invoice failed validation — the EN 16931 business rules or a channel's intake checks — at some system between you and booking. This product's conversions return failures as structured errors with rule identifiers (for example BR-16 for a line-less invoice), and a failed conversion creates an exception workspace listing each exception with its field, so corrections are made against the invoice data and re-validated.
First principle: the error is a work order
A rejection feels like a verdict on your competence; it is actually a diff between your invoice and a public rulebook. Every structured system that rejects you names the rule, and the rule names the field, and the field has a correct value you can supply. The suppliers who recover fastest treat the message as a work order with the diagnosis already written — because, uniquely in business paperwork, it literally is.
Start by capturing the rejection exactly: the rule codes, the invoice it belongs to, and the system that issued it (your client's platform, their access point, a national portal). Then reproduce the failure here before fixing anything: run the same invoice data through the converter and compare its exception list with the rejection. Where they match, you have a local laboratory for the fix — iterate here, where iterations are free, instead of against the client's intake, where each attempt costs days.
Presence and code failures: the quick wins
Missing-field rejections — no invoice number, a malformed date, an incomplete buyer address, an invoice with no lines — are the most common and the cheapest to fix. They arise from templates designed for human readers, who unconsciously forgive gaps a parser cannot. Fill the data completely; then fix the template so the gap cannot recur. A template that permits an empty postal code will produce this rejection forever, one client at a time.
Code failures are their siblings: a type code outside 380/381/384/389, a currency written as a word instead of its ISO code. Both are one-field corrections with the same moral — systems act on codes, so codes are data, not decoration. Audit your templates for every coded field and give each a constrained, correct value at the source.
VAT and identifier failures: the cross-border classics
The rejection that surprises experienced PDF-invoicers is the VAT family: a standard-rated line at 0% (BR-S-08), a zero-family category carrying a rate, a blended line mixing treatments. Underneath sits a decision error — the treatment was never consciously chosen, so the code and rate were filled by habit. Re-derive the treatment for the supply (standard, reverse charge, exempt, zero-rated), set category and rate as a pair, one treatment per line, and the whole family closes.
Identifier failures are the other classic: reverse charge across borders without the buyer's VAT identifier, or any cross-border B2B invoice without the seller's. These are not invoice defects so much as onboarding defects — the identifiers were never collected. Collect them now, record them in the client file, and add 'identifiers complete' to the gate an invoice must pass before it is issued. Future you inherits a client record that makes this rejection impossible.
Totals failures: arithmetic with a paper trail
BR-CO rejections — stated totals disagreeing with computed ones — come from one habit: typing totals. Somewhere upstream, a human or a spreadsheet wrote a payable amount that the lines no longer support: a line edited after the total was set, a discount applied in the total but not the lines, tax rounded by hand. The cure is structural, not careful: totals must be computed from lines, always, by the converter — never supplied. If your source system insists on stating totals, reconcile them against the computed figures before the invoice leaves, and treat any difference as a defect in the source document, not a rounding preference.
Allowances and charges deserve their moment here: discounts belong in the allowance fields and surcharges in the charge fields, where the arithmetic can see them — not netted silently into unit prices or appended to totals. A rejection on totals is often an allowance hiding in the wrong place, and moving it to its field fixes the arithmetic and the audit trail at once.
Resending without repeating
Resend the corrected document as a clean conversion: fix the data, re-validate until the exception list is empty, regenerate the files, and send the new pair through the client's channel with a note identifying the correction. If the original reached the client's system in any form, coordinate the resend with their finance contact — a corrected invoice (384) or a fresh submission may be what their process expects, and guessing wrong creates duplicates, which are worse than rejections.
Then close the loop properly: every rejection you ever receive should retire a defect in a template, a client record or a checklist. Keep a simple log — date, client, rule, root cause, source fix. Within a quarter the log becomes the best training document your billing process owns, and the rejection rate becomes the metric that proves the process works: not zero errors attempted, but zero errors repeated.
Rejection recovery in 5 steps
- Capture the rejection exactly: rule codes, the invoice, and the system that rejected it.
- Reproduce it in the converter and work the exception list in order: presence, codes, VAT, identifiers, totals.
- Fix the data — then fix the source (template, client record, onboarding gate) that produced the data.
- Re-validate to a clean pass and regenerate the files; never hand-edit the XML to force a pass.
- Coordinate the resend with the client's finance contact and log the root cause so the rejection cannot recur.
Worked example: one rejection, three root causes
A EUR 6,200.00 invoice to a German client is rejected with three findings: the buyer's VAT identifier is missing on a reverse-charge invoice, one line is category S at 0%, and the stated payable is EUR 40.00 adrift from the computed figure. Three findings, three different root causes: an onboarding gap (identifier never collected), a template default (rate field left at zero), and a hand-typed total carried from a quote.
Each is fixed at its own level — client record completed, template's rate made mandatory, totals henceforth computed only — and the re-validated invoice (net EUR 6,200.00, AE tax EUR 0.00, payable EUR 6,200.00, with the training line correctly re-treated or re-rated per its real supply) is resent cleanly. The log entry takes two minutes. The client's next twelve invoices, sharing the same template and record, inherit all three fixes for free.
Frequently asked questions
The rejection has a rule code I don't recognise — where do I start?
Reproduce the invoice in the converter: its exception list explains failures by field and rule, in fix order. The business rules guide decodes the families the codes belong to.
Can I just edit the XML file to fix it?
No — hand-edited files break the consistency the parse-back check guarantees, and the same defect remains in your data for the next invoice. Fix the data, regenerate, resend.
Why did the client's system reject what my PDF-era process sent for years?
Because structured intake checks mechanically what humans used to forgive: identifiers, category-rate agreement, reconciled totals. The invoice's data did not get worse; the checking got real.
Should I send a corrected invoice (384) after a rejection?
If the original was received and recorded by the client's system, possibly — coordinate with their finance contact. If it was rejected at intake, a clean resubmission of the corrected conversion is usually the path. Do not create duplicates by guessing.
How do I stop rejections recurring?
Log every rejection with its root cause and retire the cause in a template, record or checklist. Repeated rules in the log mean the source fix was skipped.
Keep going
Run the free tool this guide is about: convert an invoice · acceptance checker · pricing · about EInvoice Translator · all guides
Related guides
- How to validate an e-invoice before you send it (and read the errors)
- EN 16931 business rules explained: what BR codes mean when validation fails
- VAT categories explained: S, Z, E, AE, K, G and O on an e-invoice line
- Invoice totals and the BR-CO rules: how e-invoice arithmetic must reconcile
- Cross-border invoices: when seller and buyer VAT identifiers are required