How to Automate Data Entry into CargoWise from Freight Documents

After the native capabilities, the keying that remains usually falls into five groups:

• Document types not covered yet, such as arrival notices, delivery orders, booking instructions written in an email body, or bills of lading until their capture is available.
• Checks across documents. A bill of lading, its packing list and the commercial invoice describe the same goods; a capture that reads each one alone does not compare them.
• Matching that needs context: which open shipment a carrier's notice belongs to when its references are partial, or which of two similar organizations a consignee is.
• Documents that never reach eDocs: attachments in shared mailboxes, portal downloads, scans from a warehouse.
• Data that must also reach another system, such as a customer portal or a finance tool, in the same pass.

Measure it before deciding anything. Take one ordinary week, list every document your operators keyed, and sort them by what CargoWise covers today.

Three layers of checks, each run before a message is built:

1. Each field on its own: formats, check digits, dates read without ambiguity, weights positive and with a unit, codes present in your lists.
2. Documents against each other: packages and weights on the bill of lading against the packing list, the invoice total against its lines, the same consignee on every document of the shipment.
3. Against CargoWise: the shipment exists and is in a state that still accepts the update, the organizations resolve, and the value about to be written does not overwrite one a person entered by hand.

The tolerances are an operations decision, signed by the operations manager per document type, not a setting the vendor picks. Start strict and loosen one rule at a time with the pilot data in front of you.

Duplicates come from four places, and each needs its own key:

• The same file sent twice, for example forwarded by two colleagues: a hash of the file identifies it on arrival.
• The same document in a new file, rescanned or exported again: its type, issuer and document number identify it, and identical field values confirm the duplicate.
• A second document for a shipment that already exists: the reference lookup finds the shipment, so the workflow updates it instead of creating another one. Creating shipments from documents, if you allow it at all, is reserved for the types that open one, such as a booking instruction.
• A message sent again after a timeout: see recovery below. Never resend without checking what CargoWise received.

eAdaptor Next lets a message carry an external message ID. Give every message one, derived from the document and the version being written, so a response or a later check can be tied to it. WiseTech's flyer does not say that CargoWise ignores a second message with the same ID, so do not rely on it: the check belongs on your side.

A write fails in one of two ways, and they are handled in opposite ways. A rejection means CargoWise received the message and refused it, for a missing code, a record in the wrong state or a value it does not accept: retrying sends the same refusal, so the item goes to review with CargoWise's own error text. A transport failure means the message may or may not have arrived (a timeout, an unreachable endpoint, refused credentials): the item stays in the queue and is retried, after checking.

• Keep an outbox: every message is stored with its status (built, sent, acknowledged, rejected, waiting to retry) before it is sent.
• Retry transport failures with increasing intervals and a ceiling; alert a named person when the ceiling is reached.
• After a timeout, check before resending: look up the record by its references, or match the outbound response by message ID, and resend only what did not arrive, with the same ID.
• Pause all sending on an authentication error. The queue keeps the items; nothing is dropped.
• Reconcile daily: documents received against documents with a final status. The balance must be zero.

The Connector Manager keeps transaction history and error messages on the CargoWise side. Your workflow's own log should still be complete on its own, because the question after an incident is "which documents are affected", and that starts from the document, not from the message.

Bills of lading get amended, invoices get corrected, arrival notices get reissued with new charges. Treat each as a version of the same document, keyed on type, issuer and number:

• Keep every version unchanged, with its hash and arrival time. Never overwrite the original file.
• Compare the new version with the previous one field by field, and update only the fields that changed.
• If a person edited one of those fields in CargoWise after the first write, hold the update: the operator decides which value stands.
• File each version in the shipment's eDocs under the right document type, so the operator sees the history next to the data. How the file reaches eDocs (in the message or by a separate route) is agreed with your administrator.
• Late documents for a shipment already closed or invoiced go to review, whatever they contain.

The build is the smaller part. What keeps the workflow useful is what happens in the months after:

When a supplier changes its format

A shipper changes its invoice template, a carrier reissues its arrival notice layout. The first sign is a rise in held items for that issuer. Keep a stored set of real documents per issuer and type; every change to the extraction or the mapping is run against that set, and the results compared, before it is released. Agree in writing who makes the change and how fast.

When CargoWise changes

New releases, a change to a message format, or a move from SOAP to eAdaptor Next can affect the messages. Ask your administrator to warn the team running the workflow before an update, and re-run the stored set against the non-production instance when one exists.

When your own lists change

New charge codes, merged organizations, new eDocs types. The mapping reads these lists from CargoWise rather than from a copy made once, or the first new code sends every document that uses it to review.

What to watch every week

• Held items by document type and issuer, and the median time to a decision.
• Rejections by CargoWise, by error, which point to a mapping or a code list to fix.
• Fields corrected after review: a correction made often is a rule to change.
• Items without a final status at the end of the day.

Compare options on everything they cost, not only the subscription:

• The native option: whatever WiseTech charges for the portals on your license, and the review time they leave.
• An outside workflow: the build and the subscription, the administrator's time to set up eAdaptor and the codes, any WiseTech fee for a non-production instance, and the review time it leaves.
• On both sides: who maintains the mapping when formats change, and what an error that reaches a customs filing or an invoice costs you.

Then run a pilot that writes nothing to production until it has earned it: read only first, with operators keying as usual and the results compared field by field; then writes to the non-production instance; then one document type in production. The general method is in the document automation pilot guide. The pass marks specific to CargoWise:

Mirage builds and runs this kind of workflow: it reads the documents your team keys, runs the checks above and writes into the system you already run. For CargoWise it connects through eAdaptor, with credentials your administrator issues and can revoke. Mirage holds no WiseTech partnership, certification or listing and has not published a CargoWise deployment; its offer, and the questions to put to any vendor, are on the CargoWise integration page.

Its published integrations are with other systems. At Transwin, a Moroccan logistics operator, Mirage reads freight and customs documents and writes them into Transwin's TMS; the case study reports 70% less manual data entry on freight documents, for that deployment. At OFRET Groupe, Mirage connected to the Logitude World API and started with freight document extraction before moving to quote requests. Neither runs on CargoWise. For other systems, see TMS data entry automation.