The short answer
A rejected Peppol invoice must not be resent blindly. First record the status, timestamp, document identifier and validation message. Then classify the cause: recipient not found, missing mandatory data, calculation rule, duplicate, transport incident or business dispute. Correct the source in Odoo, regenerate the document if needed and retain evidence of the new submission and acceptance.
- Identify the exact state before acting
- Correct the source, not just the file
- Separate technical matters and business decisions
- Prevent duplicates through evidence
- Measure recurring causes
What does ‘rejected’ actually mean?
‘Rejection’ covers several states. A document may fail before reaching the access point, be rejected by a format rule, arrive technically but not enter the customer’s software, or be disputed after receipt for a commercial reason. These situations require different evidence and corrections. Before acting, ask which system produced the status and which point in the flow it refers to.
Since 1 January 2026, Belgian VAT-liable businesses must use structured electronic invoices for their B2B transactions concerned. The federal portal states that a PDF sent by email is insufficient and exchanges take place through Peppol. This makes status interpretation operational: ‘sent’ does not automatically mean ‘delivered’, and ‘delivered’ means neither accounting validation nor agreement on the service.
Collect evidence before changing Odoo
Start by preserving the facts. Record the internal number, electronic identifier, issuing company, recipient, access point or provider, sending date, exact status and complete message. Export the XML or available log without changing it. A screenshot is useful, but machine-readable text and the timestamp allow more reliable investigation with the vendor or provider.
Avoid deleting the invoice, changing several fields and resending. This erases the cause and can create a duplicate if the first document was ultimately transmitted. In a multi-company organisation, also record the database, environment, journal and user. Teams operating web applications and business tools can then link the exception to the right business data rather than treating every incident as a generic Odoo defect.
Classify the exception into six families
Good classification reduces processing time. The first family concerns identification: Peppol participant not found, VAT number, identifier scheme or inconsistent address. The second covers mandatory fields and references. The third groups calculations, taxes, rounding and units. The fourth concerns a duplicate or lifecycle inconsistency. The fifth concerns transport, the access point or unavailability. The sixth is a business refusal: unknown order, disputed price or unrecognised service.
The category determines ownership. Finance rarely fixes a network incident; IT does not decide whether a credit note is legally required; sales does not change a tax identifier without approval. Add category, severity, owner and deadline to the ticket. An unclear message can remain ‘to be classified’, but must not automatically become a ‘Peppol problem’ without evidence.
Compare statuses, evidence and responses
Read the status alongside its evidence. A validation rejection calls for document correction; a transport incident may require a controlled retry; a business dispute requires a person, not just a button. The table below provides a guide, as exact labels vary by Odoo version, localisation and provider.
Do not treat a positive notification as universal approval. Technical acceptance confirms a given network or validation stage, while the recipient organisation may still apply internal checks. Conversely, no screen update does not establish that the document is lost: cross-check the Odoo log, provider portal and, if needed, the recipient’s acknowledgement.
| Type | Evidence | Appropriate response | Avoid |
|---|---|---|---|
| Structural validation | Rule code, field or path | Correct source data, then regenerate | Force a total or edit the XML |
| Recipient not found | Participant or identifier not resolved | Check the entity and registration | Try random identifiers |
| Transport blocked | Provider log and timestamp | Escalate, then retry in a controlled way | Create multiple numbers |
| Possible duplicate | Identifier already seen or ambiguous acknowledgement | Freeze resubmission and confirm the initial state | Resend automatically |
| Business refusal | Customer reason or order discrepancy | Have finance/purchasing/sales decide | Classify it as an IT incident |
Improve master data reliability
Most lasting corrections are made in master data, not the final XML. Check the legal name, company or VAT number, address, country, participant identifier, payment details and customer-required values. Confirm who may edit these fields and based on which source. A one-off invoice correction must not hide an incorrect partner record that will cause another rejection tomorrow.
For groups, associations with distinct activities or multi-site organisations, do not assume one identifier covers every entity. Distinguish legal entity, establishment, delivery address, billing address and contact. Document the format requested by customers requiring an order or contract reference. Finally, test synchronisation with CRM, e-commerce or accounting imports: correct data in Odoo may be overwritten by an integration.
Understand validation rules
Peppol BIS Billing applies structural and arithmetic rules. A line must remain consistent across quantity, price, discount and net amount; taxes and totals must reconcile under the applicable rules. OpenPeppol documentation publishes these checks in a verifiable form. The rejection message often contains a code and path pointing to the data concerned, even if it remains difficult for a finance user to read.
Do not manually change the total to force acceptance. Find the source: an unusual unit, decimal price, discount, negative quantity, incorrect VAT category or rounding from a connector. Reproduce the calculation with source data and keep a minimal example. If the problem depends on a version or module, test the correction on a representative copy before production.
Correct and resend without creating duplicates
The safe sequence involves five decisions: confirm the first submission’s state, classify the cause, correct the source, generate a new document under the chosen process, then check the acknowledgement. If the invoice was never delivered and remains editable under your procedure, correction and reissue may be appropriate. If it was received or posted, finance must decide on cancellation, a credit note or a corrective document.
Avoid parallel numbers created just to ‘try’. Link every new submission to the incident and original document. State in the ticket who confirmed there was no duplicate and which evidence closes the action. Screen labels vary: the runbook should describe expected outcomes, not a fragile sequence of clicks. GVISION also recommends a four-eyes check for sensitive amounts or tax data corrections.
Handle incoming exceptions too
Received invoices may be technically valid yet fail in the internal process: unrecognised supplier, missing order, unknown cost centre or amount discrepancy. Separate Peppol checks from business matching. The first checks structure and transmission; the second decides whether the organisation can record, allocate and approve the expense.
Create an incoming exception queue with reason, owner and deadline. Finance handles accounting data, purchasing the order, the requester the service and IT the integration. Do not put every problematic invoice in a personal mailbox. A shared queue or tracking board prevents absence blocking payment. Replies to suppliers should cite a usable reference without unnecessarily exposing personal data.
Automate without hiding errors
Useful automation detects, routes and documents; it does not turn every error into an automatic resubmission. Trigger an alert when a status remains stuck, the same code recurs or exception volume exceeds a defined threshold. Enrich the ticket with company, partner, document identifier, likely category and evidence link. No secret or complete XML should be shared through an overly broad channel.
An approach tobusiness process automation can synchronise Odoo, a ticketing tool and a dashboard. GVISION favours explainable rules: every automatic action has a condition, log and escalation route. Start with repetitive, low-risk cases, then measure false positives before automating data correction or reissue.
Assign roles and escalation levels
A robust process names four roles. The business owner decides on the reality of the transaction. Finance validates accounting and tax treatment. The application administrator checks configuration, data and integrations. The provider or access point intervenes when evidence indicates a transport or technical compliance problem. For each error family, specify who analyses, corrects, approves and informs the customer.
Also define priorities. One invoice without an immediate deadline has a different impact from all outgoing invoices being blocked at period-end. Use volume, value, due date, customer and duplication risk. An escalation includes a minimal reproducible case, never just ‘Peppol does not work’. This speeds external help and avoids exchanges of conflicting screenshots.
Test before and after every change
Build a small scenario catalogue: simple invoice, multiple rates, discount, credit note, order reference, relevant foreign customer, partner not found and rounding boundary value. Scenarios should reflect actual flows without using production personal data. When the provider offers a test environment, check what it actually simulates; test success does not guarantee registration of the real recipient.
After updating Odoo, a module or a connector, rerun critical scenarios. Compare the resulting document, statuses and matching. Keep the version, date and result. Test permissions too: a user should see enough information to act without being able to change tax data or resend an invoice outside their role.
Manage exceptions with a few indicators
Track rejections per hundred documents, breakdown by cause, median classification time, resolution time, reopening rate and duplicates avoided or detected. These indicators improve data and process, not punish teams. An overall rate may hide that one connector, customer or entity causes most incidents.
Review trends weekly at the start, then adapt the frequency. Recurring errors should lead to action on the cause: entry rules, pre-send checks, integration correction or training. Also keep a list of statuses without a clear outcome. It informs documentation and provider requests. Do not publish a target turnaround without real support capacity and distinguishing technical incidents from business decisions.
Implement the runbook in 30 days
In week one, inventory statuses, available evidence, frequent partners and roles. In week two, create the cause taxonomy and ticket template. In week three, test five scenarios and document correction, resubmission and escalation. In week four, enable simple alerts, train finance and support, then measure the first exceptions. A short runbook that is used is worth more than an exhaustive manual nobody consults.
Common mistakes are known: resending before checking, editing XML instead of the source, confusing delivery and approval, sharing sensitive data in a public channel, closing without evidence or automating risky reissue. Review the process after each major incident and software change. The federal portal announced the end of general tolerance on 7 April 2026; exceptional technical difficulties are reviewed individually, increasing the value of dated evidence.
Frequently asked questions
Is a ‘sent’ invoice necessarily received?
No. Check transport status and the available acknowledgement; initiating a send does not establish delivery.
Can you simply resend the same invoice?
Only after establishing the first submission’s state and duplication risk. The procedure also depends on accounting treatment already performed.
Should you edit the XML directly?
Not in normal operation. Correct the source data and let the application regenerate a traceable document.
Does a Peppol rejection mean Odoo is down?
No. The cause may be partner data, a validation rule, a transport incident or a business requirement.
What should you keep in the ticket?
Identifiers, timestamps, exact status, code and message, environment, action, approval and final evidence.
When should you issue a credit note?
The decision depends on the document’s legal and accounting status. Finance or the accountant must approve it.
Can you automate resubmission?
Only for precisely defined, low-risk cases with duplicate checks, logging and escalation.
Conclusion: turn rejection into a controlled process
A well-handled rejection improves data, controls or integration. The expected outcome is not just a ‘green’ invoice, but understandable evidence, an owner and prevention of recurrence. To turn this runbook into Odoo flows, ticketing and responsibilities suited to your organisation, you can scope your exception handling with GVISION.
To go further, explore our support: Our Odoo support.
Official sources
Published on . Updated on .
Let’s talk about your project
Want to apply this advice in your organisation? Let's review your needs and next steps together.
Talk to GVISION



