Plan and verify a Visma order export EN
Define the Kaduno → PullPush → Visma handover, compare fields and handle uncertain delivery without duplicate orders.
This implementation guide describes an order-export pattern: Kaduno order → configured PullPush flow → Visma sales order. It is not a one-click setup promise or confirmation that your connector has passed acceptance. PullPush is in beta. Actual mappings, credentials, enabled flows and responsibility for invoicing must be agreed for the installation.
This guide deliberately contains no customer orders, negotiated prices, personal contact details or production connection identifiers. Use approved synthetic data for a controlled test. Source-code support is not evidence that a feature is currently enabled in your store.
1. Agree the handover before enabling a flow
Record these decisions with the integration owner:
| Decision | What to confirm |
|---|---|
| Store and destination | Correct Kaduno store, PullPush workspace/flow and Visma company/environment |
| Customer identity | Which system owns customer numbers and how the correct company is matched |
| Billing versus delivery | Whether the existing ERP customer card owns invoice details; how each order's delivery address and recipient are mapped |
| Financial responsibility | Which system creates and sends invoices, and how card, invoice and provider-managed payment methods are represented |
| Product and shipping mapping | Item identifiers, units, freight lines, shipping codes and pickup versus delivery |
| Data ownership | Which fields may be changed by each system, and what is intentionally not synchronised |
| Operations | Who reviews failed/skipped deliveries, who may retry, and how destination records are checked first |
Use separate credentials for the agreed environment with the minimum required permissions. Keep tokens in approved secret storage, not the implementation record. A saved credential does not prove a working connection. Do not switch a production flow on just to see whether the mapping works.
2. Separate an order from its invoice
When Kaduno is configured for external ERP invoicing, the order is exported and the ERP creates the invoice; Kaduno's invoice-trigger logic does not create a second invoice for that order. This is different from the accounting-hub flow described in Norwegian. Confirm the actual provider configuration before relying on this boundary.
A sales order in Visma is not evidence of payment settlement, shipment, an issued
invoice or reconciliation. A card authorisation is not the same as captured
payment. The export contract distinguishes settlement states such as authorized,
paid and open; the destination mapping must preserve the intended meaning.
Do not infer settlement from the payment-method label alone.
3. Compare the source and destination field by field
Use one agreed test scenario at a time, with no real customer charge, shipment or invoice unless separately authorised. Retain a redacted comparison, not a copy of personal data in a public ticket.
- Traceability: the source order number and the destination sales-order identifier can be correlated. A buyer reference and an order message are separate fields; verify both when required.
- Company and person: the buying company's identity is not the same as the person placing the order. Verify the destination's customer, contact and attention/recipient fields separately.
- Addresses: invoice address, delivery address and pickup location have different meanings. An existing invoice address on an ERP customer card is not proof that this order's delivery address is right.
- Payment: verify the mapped method and settlement meaning for each supported payment scenario. Do not reuse one customer's ERP method codes elsewhere.
- Lines and totals: compare item/variant, unit, quantity, discount, tax basis, freight, VAT and total in the same currency and minor-unit convention. A correct grand total can still hide mixed inclusive/exclusive VAT rows.
- Free text: if a message is required, first check that it was saved on the Kaduno order, then check its destination field. Do not blame the connector for a value absent at source.
The API exposes fields that a connector can use; that does not mean every flow maps every field. Review the configured mapping and the actual received record. A record repaired manually after export is not evidence of an automatic mapping.
4. Read delivery state correctly
For a configured PullPush order-export flow, Kaduno derives the status from the order's recorded delivery attempts. The newest matching attempt determines the state shown; inspect the history and destination when the outcome is uncertain.
| State | Meaning and next check |
|---|---|
| Not configured | The integration is not configured for this context; check setup before expecting export |
none | No matching order-export flow or no recorded event for this order; absence of evidence is not success |
pending | The order has been seen but no matching delivery attempt is recorded yet; inspect the flow/queue rather than creating another order |
delivered | The latest matching attempt reports delivery; compare the destination identifier and business fields, not just the status label |
failed | Read the recorded error and reason, correct the cause and determine whether anything was created remotely before retrying |
skipped | Inspect the reason and flow rules; do not treat a skipped order as delivered |
If a timeout or missing callback leaves the outcome unclear, search for the existing destination record using the agreed correlation before retrying. Never create another order merely to clear a warning. Do not promise exactly-once execution from a successful status alone.
5. Treat return flows as separate acceptance work
Order export does not prove stock, shipment or invoice return flows. In the
shipment-ingest contract, a response with HTTP 200 can contain failed groups;
callers must inspect each outcome. Live dispatch can trigger payment/invoice
lifecycle effects and is not a harmless connectivity check. Agree and test it
separately with the people responsible for fulfilment and finance.
An address model or mapping that is still under discussion is not a final synchronisation contract. Define its owner, expected changes and conflict handling before rollout rather than assuming full bidirectional synchronisation.
Acceptance record
For each scenario, record: environment/version, agreed mapping, synthetic input, expected destination fields, observed result, correlation, reviewer and remaining gaps. Mark passed, failed, not tested and manually corrected separately. Agree test-data cleanup in both systems; cancelling in one is not proof that the other has been cancelled.
Continue with the integration directory, read-only API guide and local webhook exercise.